Connecting the Dots: From an Easy Method to Computerized Species Determination

Transcription

1 insects Article Connecting Dots: From an Easy Method to Computerized Species Determination Senta Niederegger 1, *, Klaus-Peter Döge 2, Marcus Peter 2, Tobias Eickhölter 2 and Gita Mall 1 1 Institute Legal Medicine, University Hospital Jena, Am Klinikum 1, Jena, Germany; Gita.Mall@med.uni-jena.de 2 Department Electrical Engineering and Information Technology, University Applied Sciences Jena, Carl-Zeiss-Promenade 2, Jena, Germany; Klaus-Peter.Doege@eah-jena.de (K.-P.D.); Marcus.Peter@stud.eah-jena.de (M.P.); eickhoelter.tobias@googl .com (T.E.) * Correspondence: senta.niederegger@med.uni-jena.de; Tel.: Academic Editors: David Rivers and John R. Wallace Received: 8 March 2017; Accepted: 12 May 2017; Published: 18 May 2017 Abstract: Differences in growth rate forensically important dipteran larvae make species determination an essential requisite for an accurate estimation time since colonization body. Interspecific morphological similarities, however, complicate species determination. Muscle attachment site (MAS) patterns on inside cuticula fly larvae are species specific and grow proportionally with animal. The patterns can refore be used for species identification, as well as age estimation in forensically important dipteran larvae. Additionally, in species where determination has proven to be difficult even when employing genetic methods this easy and cheap method can be successfully applied. The method was validated for a number Calliphoridae, as well as Sarcophagidae; for Piophilidae species, however, method proved to be inapt. The aim this article is to assess utility MAS method for applications in forensic entomology. Furrmore, authors are currently engineering automation for pattern acquisition in order to expand scope method. Automation is also required for fast and reasonable application MAS for species determination. Using filters on digital microscope pictures and cross-correlating m within ir frequency range allows for a calculation correlation coefficients. Such pattern recognition permits an automatic comparison one larva with a database MAS reference patterns in order to find correct, or at least most likely, species. This facilitates species determination in immature stages forensically important flies and economizes time investment, as rearing to adult flies will no longer be required. Keywords: forensic entomology; muscle attachment sites; species determination; image processing; correlation; knowledge base 1. Introduction Blowflies ten start biological clock for postmortem interval (PMI) calculation in forensic entomology. They might approach a body within first hours after death [1 3] for oviposition. The age oldest blowfly found on a body or carcass is refore a good estimate for minimum postmortem interval or minimum exposure time [4 6]. Laboratory experiments growing larvae known species under controlled temperature regimes showed that age correlates to body length animals [7 11]. However, even though larvae different species look extremely similar, y might grow to different body lengths in same time and under identical conditions [11]. Species and age determination can refore be a great challenge when a conglomerate unidentified larvae is collected from a body in field. Insects 2017, 8, 52; doi: /insects

2 Insects 2017, 8, AInsects common 2017, 8, 52 method for species determination is to allow collected larvae to pupate and 2 16 hatch, and to determine species by morphological features imago [3,12]. This is very time-consuming A common method for species determination is to allow collected larvae to pupate and hatch, and requires living larvae. Or methods based on light microscopy rely on morphological and to determine species by morphological features imago [3,12]. This is very characteristics larvae, like shape cephalopharyngeal skeleton (CPS), spiracles, or time-consuming and requires living larvae. Or methods based on light microscopy rely on spinebands morphological [13 22]. characteristics These methods can larvae, be like hindered shape by fragmentation cephalopharyngeal or discoloration skeleton (CPS), specimen spiracles, by poor or spinebands preservation. [13 22]. Technically These methods advancedcan approaches be hindered aimby toward fragmentation DNA-based or species identification discoloration using molecular specimen biological by poor preservation. methods [23 25] Technically or electron advanced microscopy approaches [26,27]. aim toward Requiring cost-intensive DNA-based andspecies specialized identification equipment, using se molecular methodsbiological might notmethods be applicable [23 25] in every or electron laboratory. microscopy [26,27]. Requiring cost-intensive and specialized equipment, se methods might not be 2. Approach applicable from in every Biological laboratory. Perspective A2. fast Approach and easy from method, Biological suited Perspective for all larval instars and even applicable to a fragmented or discolored specimen, was developed in 2012: While attempting to locate trichoid sensilla, round A fast and easy method, suited for all larval instars and even applicable to a fragmented or structures were detected on inside cuticula a number brachyceran larvae. Furr discolored specimen, was developed in 2012: While attempting to locate trichoid sensilla, round investigation structures revealed were detected that y on were inside attachment cuticula sites for transversal a number body brachyceran wall muscles. larvae. Furr The thoracic segments, investigation as well as revealed first that andy last were abdominal attachment segments, sites for eachtransversal possess a body unique wall muscle muscles. arrangement, The whilethoracic remaining segments, abdominal as well as segments first and share last anabdominal identical pattern segments, [28,29]. each possess The body a unique wall musculature muscle many arrangement, larvae is composed while remaining longitudinal, abdominal diagonal, segments andshare transversal identical muscles pattern (Figure [28,29]. 1a). The Larval body flies lack circumferential wall musculature muscles many aslarvae antagonists composed to longitudinal, muscles diagonal, [30]. and The transversal vertical displacement muscles (Figure segments 1a). Larval is refore flies lack probably circumferential supported muscles by as antagonists transversal to muscles longitudinal which are muscles located [30]. in The vertical displacement segments is refore probably supported by transversal external layer, where y form blocks in each segment and are directly attached to cuticula [31] muscles which are located in external layer, where y form blocks in each segment and are (Figure 1b). directly attached to cuticula [31] (Figure 1b). Figure 1. (a) Opened larva with complete musculature; (b) The longitudinal muscles were removed Figure to 1. reveal (a) Opened underlying larva with transversal complete muscle musculature; groups. (b) The longitudinal muscles were removed to reveal underlying transversal muscle groups Species and Age Determination 2.1. SpeciesThe andmuscle Age Determination attachment sites (MAS) se transversal muscles could easily be visualized by The removing muscle attachment muscles from sites cuticle (MAS) and subsequent se transversal staining with muscles Commassie couldbrilliant easily blue. be visualized The analysis MAS revealed that structures are arranged in rows and form distinct patterns (Figure 2a). by removing muscles from cuticle and subsequent staining with Commassie brilliant blue. Ten larvae each species were prepared in this fashion and pictures were recorded. A small number The analysis MAS revealed that structures are arranged in rows and form distinct patterns row subsets located at center last two thoracic and first abdominal segment (Segments 2 (Figure 4) were 2a). Ten manually larvae charted. each Correlating species were rows were prepared stacked inand thiscondensed fashion and into row pictures patterns (Figure were recorded. 2b) A small and number described for rowa few subsets forensically locatedimportant center species blowflies last two (Diptera: thoracic Calliphoridae) and first Calliphora abdominal segment vomitoria, (Segments C. vicina, 2 4) Lucilia were sericata manually [32], Protophormia charted. terraenovae Correlating [33], and rows Phormia wereregina stacked [34]. and The results condensed into row confirmed patterns interspecific (Figurevariation 2b) andwhich described allowed for a few species forensically differentiation important se five species. There blowflies (Diptera: can be Calliphoridae) variation in Calliphora number vomitoria, attachment C. points vicina, composing Lucilia sericata a row [32], in larvae Protophormia same terraenovae age and [33], and Phormia regina [34]. The results confirmed interspecific variation which allowed for species differentiation se five species. There can be variation in number attachment points

3 Insects 2017, 8, composing a row Insects 2017, 8, 52 in larvae same age and species. Such variability is common in 3biological 16 systems and also present in or morphological characteristics, such as ratio length species. is common in biological systems and tip also[35] present in length or morphological Insects 2017,Such 8, 52 variability 3 16 mouth hook base and distance between base and or and width characteristics, such as ratio length mouth hook base and distance between base posterior spiracle discs [36]. The reason for divergence in MAS numbers remains unclear, but does species. variability is common systems and also present in or morphological and Such tip [35] or length and widthinbiological posterior spiracle discs [36]. The reason for divergence not affect usability method. characteristics, such as unclear, ratio affect mouth hook base and distance between base in MAS numbers remains butlength does not usability method. and tip [35] or length and width posterior spiracle discs [36]. The reason for divergence in MAS numbers remains unclear, but does not affect usability method. Figure 2. (a) All muscles were removed from transparent cuticle, and attachment sites were Figure 2. (a)with AllCoomassie muscles were removed fromare laterally transparent cuticle, and attachment sites were colored brilliant blue. Patterns symmetrical. Investigated muscle groups colored with 2A, Coomassie brilliant Patterns are laterally symmetrical. Investigated muscle groups labelled 3A, 3B, and 4A, 4B,blue. according to segments; (b) Patterns ten larvae same species Figure 2. (a) All muscles were removed from transparent cuticle, and attachment sites were labelled 2A, 3A, 3B, with and 4A, 4B, according to segments; (b) Patterns dots. ten larvae same species photographed, attachment sites marked with semitransparent Resulting rows were colored with Coomassie brilliant blue. Patterns are laterally symmetrical. Investigated muscle groups stacked andwith areasattachment with a highsites degree overlap marked in black.dots. Resulting rows were stacked photographed, marked withare semitransparent labelled 2A, 3A, 3B, and 4A, 4B, according to segments; (b) Patterns ten larvae same species and areas with a high degree overlap are marked in semitransparent black. photographed, with attachment sites marked with dots. Resulting rows were The observation larvae in different developmental stages furr showed that basic stacked and areas with a high degree overlap are marked in black. central muscle attachment site patterns are conserved throughout development. The length each The larvae in different developmental stages furr showed that thus basic central row,observation however, increases with body length maggot (Figure 3). The method could not The observation larvaeare in different developmental stages furr showed that basic row, muscle attachment site patterns conserved throughout development. The length each only be used for species determination, but also for age determination [33]. central muscle attachment site patterns are conserved throughout development. The length each however, increases with body length maggot (Figure 3). The method could thus not only be row, however, increases with body length maggot (Figure 3). The method could thus not used for species determination, but also for age determination [33]. only be used for species determination, but also for age determination [33]. Figure 3. Pattern consistency during growth C. vomitoria. Figure 3. Pattern consistency during growth C. vomitoria. Figure 3. Pattern consistency during growth C. vomitoria.

4 Insects 2017, 8, Insects 2017, 8, Insects Genus 2017, Lucilia 8, Genus Lucilia 2.2. ToGenus furr Lucilia challenge method, six closely related and forensically relevant species genus To furr challenge method, six closely related and forensically relevant species Lucilia were tested [37]. In order to achieve reliable results, set analyzed MAS patterns had genus To Lucilia furr were challenge tested [37]. method, In order six to achieve closely reliable related results, and forensically set analyzed relevant species MAS patterns to be enlarged. While isolated central patterns yielded adequate results in first experiments [32], genus had to Lucilia be enlarged. were tested While [37]. isolated In order central to achieve patterns reliable yielded results, adequate results set analyzed in first MAS experiments patterns it became necessary to include all rows present on segments two, three, and four. The labeling MAS had [32], to it became enlarged. necessary While to isolated include central all rows patterns present yielded on segments adequate two, results three, in and four. first experiments The labeling rows [32], MAS was it became changed rows was necessary accordingly changed to accordingly include (Figure all 4, rows (Figure shown present 4, for shown Calliphora on segments for Calliphora vicina). two, vicina). three, and four. The labeling MAS rows was changed accordingly (Figure 4, shown for Calliphora vicina). Figure 4. New labeling in segment 4 shown for Calliphora vicina. Figure 4. New labeling in segment 4 shown for Calliphora vicina. Figure 4. New labeling in segment 4 shown for Calliphora vicina. A joint genus pattern (Figure 5) could be found by combining pictures all six Lucilia species tested AA joint (L. joint ampullacea, genus pattern L. caesar, (Figure L. illustris, 5) could L. richardsi, be found L. by sericata, combining and L. pictures silvarum). all Each all six six species Lucilia differed species species tested tested from (L. (L. genus ampullacea, pattern L. L. caesar, in a characteristic L. L. illustris, L. way. richardsi, This L. allowed sericata, for and L. L. silvarum). composition Each Each species distinct differed species from from patterns for genus each pattern in in six aa species. characteristic For L. way. illustris This and allowed L. caesar where for composition species determination distinct species patterns larvae for was for reported each to six be six difficult species. [38 40] For L. illustris or impossible and L. [41], caesar where even with genetic species methods determination [42] MAS larvae larvae patterns was was were reported distinctive to to be be [37]. difficult [38 40] or impossible [41], even with genetic methods [42] MAS patterns were were distinctive [37]. Figure 5. Genus pattern for Lucilia composed 76 individual MAS patterns. Figure Genus pattern for Lucilia composed 76 individual MAS patterns Genus Sarcophaga Genus Genus Flesh Sarcophaga Sarcophaga fly (Sarcophaginae) larvae, including genus Sarcophaga, can easily be recognized from large and deep spiracular cavity on last segment [43]. Species identification, however, is Flesh Flesh fly fly (Sarcophaginae) (Sarcophaginae) larvae, larvae, including including genus genus Sarcophaga, Sarcophaga, can can easily easily be be recognized recognized from from much large more and difficult. deep spiracular Therefore, cavity next on step last in segment development [43]. Species identification, method was designed however, for is large and deep spiracular cavity on last segment [43]. Species identification, however, is much more much five species more difficult. Therefore, genus Sarcophaga next step from in five development different subgenera. method Sarcophaga was designed albiceps, for S. difficult. Therefore, next step in development method was designed for five species five argyrostoma, species S. caerulescens, genus S. Sarcophaga melanura, and from S. similis five were different previously subgenera. found Sarcophaga to be frequent albiceps, visitors S. genus Sarcophaga from five different subgenera. Sarcophaga albiceps, S. argyrostoma, S. caerulescens, argyrostoma, and successful S. caerulescens, colonizers S. large melanura, animal and carrion S. similis and were human previously corpses [44 50]. found to As be for frequent six visitors Lucilia S. melanura, and S. similis were previously found to be frequent visitors and successful colonizers and species, successful a genus colonizers pattern could large be animal generated carrion from and se human five corpses Sarcophaga [44 50]. species As for (Figure six Lucilia 6) and large species, species-specific animal a genus carrion pattern and human variations could corpses be were generated found [44 50]. [51]. from As The for se previously five six Lucilia Sarcophaga used species, set species MAS a genus did (Figure pattern not have 6) and could to be species-specific generated from pattern se variations five Sarcophaga were found species [51]. (Figure The previously 6) and species-specific used set MAS pattern did not variations have to

5 Insects Insects 2017, 2017, 8, 528, Insects 2017, 8, be extended. The genus pattern for Sarcophaga species is distinctively different from Lucilia were be genus found extended. pattern. [51]. The This genus previously indicates pattern used presence for set Sarcophaga MAS variations did species notbetween have is distinctively togenera be extended. and different families Thefrom genus which pattern promise Lucilia for Sarcophaga genus to be very pattern. species useful This for distinctively indicates MAS method. different presence from variations Lucilia between genusgenera pattern. and This families indicates which promise presence variations to be very useful between for genera MAS and method. families which promise to be very useful for MAS method. Figure 6. Genus pattern for Sarcophaga composed 50 individual MAS patterns. Figure 6. Genus pattern for Sarcophaga composed 50 individual MAS patterns. Figure 6. Genus pattern for Sarcophaga composed 50 individual MAS patterns Family Piophilini Family Family Piophilini Piophilini Preliminary analyses Piophilini species showed that ir muscle equipment is clearly different Preliminary Preliminary from analyses analyses Calliphoridae Piophilini Piophilini and species Sarcophagidae species showed showed that (Figure ir that 7). muscle ir The muscle very equipment similar equipment is species clearly is patterns clearly different from different observed Calliphoridae from between Calliphoridae and two Sarcophagidae closely and related Sarcophagidae Piophilina (Figure 7). (Figure genera The very and 7). The similar a more very distantly species similar patterns species related genus patterns observed between observed Thyreophorina, between two closely however, related two suggest closely Piophilina that related genera Piophilina larval and MAS agenera more patterns distantly and may a more be related highly distantly genus conserved related Thyreophorina, among genus however, Thyreophorina, group [52]. suggest Sufficiently however, that conspicuous suggest larval MAS that variations patterns larval for MAS may reliable patterns be highly species may conserved differentiation be highly among conserved were not among group visible [52]. to Sufficiently group naked [52]. conspicuous eye. Sufficiently variations conspicuous forvariations reliable species for reliable differentiation species differentiation were not visible were tonot visible nakedto eye. naked eye. Figure Preliminary family pattern for Piophilini. Figure 7. Preliminary family pattern for Piophilini Next Next Steps Steps 2.5. Next Steps The The main main aim aim method method is is fast fast and and reliable reliable species species identification identification in in single, single, unknown unknown larvae. larvae. A vast The pool main aim data, as well method as reference is fast and genus reliable and species identification patterns, is an in essential single, unknown requirement larvae. for A vast pool data, as well as reference genus and species patterns, is an essential requirement for A comparison. vast pool It data, is refore as well as vital reference importance genus and to optimize species patterns, process is an essential pattern requirement acquisition. The for comparison. It is refore vital importance to optimize process pattern acquisition. The manual comparison. manual charting It is refore patterns is vital very time importance consuming. to optimize The manual process generation pattern reference acquisition. patterns The for charting patterns is very time consuming. The manual generation reference patterns for genera manual genera and charting species patterns can never is very be objective. time consuming. An automation The manual generation process is refore reference required patterns for for a and species can never be objective. An automation process is refore required for a successful genera successful and extension species can never be knowledge objective. base An automation method. process It could is furrmore refore required permit for a extension knowledge base method. It could furrmore permit distinction patterns successful distinction extension patterns with only knowledge marginal base differences. A method. large data It base, could on furrmore or hand, permit is reliant with only marginal differences. A large data base, on or hand, is reliant on efficient comparison distinction patterns with only marginal differences. A large data base, on or hand, is reliant methods to achieve useful results. The comparison patterns for species determination necessitates

6 Insects 2017, 8, on efficient comparison methods to achieve useful results. The comparison patterns for species adetermination degree expertise necessitates and using a degree an extensive expertise database and would using an be very extensive time-consuming. database would Customized be very stware time-consuming. is refore Customized only stware option tois allow refore a widespread only option andto successful allow for a application widespread and MAS successful method. application MAS method. 3. Problem Analysis from Image Processing Perspective In following, image processing will will be be examined as as a potential method refinement. Digital image image processing processing has has continuously continuously developed developed over over last decades. last decades. However, However, for this method, for this it method, was necessary it was necessary to test to general test applicability general applicability before process before process development. development. This is mainly This is mainly due to due variability to variability image information image information and was and done was indone a series in a series example example images images different different species. Images species. Images larvae belonging larvae belonging to instarto L3instar were L3 exclusively were exclusively used for used all for all following following tests. tests. The evaluation showed that patterns within same segments a species are recognizable, but variable. Neir number spots nor general shape is predictable, which is illustrated by rectangle in Figure 8. For a human observer, differentiation species is is still possible [32,37]. For an automatized procedure, procedure, variability variabilitypresents presentsa achallenge due due to to missing missing contextualization. contextualization. In In addition addition to to variability variability patterns, patterns, differences differencesin in sizes sizes larvae larvaeneed needto tobe betaken into account in automated processes. Figure 8. Examples image information variability in fourth segment Lucilia ampullacea. Figure 8. Examples image information variability in fourth segment Lucilia ampullacea. Variability individual patterns (rectangle), reflections (circle), and variation resolutions (arrow) Variability individual patterns (rectangle), reflections (circle), and variation resolutions (arrow) are shown. are shown. Furrmore, transforming three-dimensional objects into two-dimensional images leads to differences Furrmore, in imagesharpness, transformingwhich three-dimensional is indicated by objects arrow into in two-dimensional Figure 8. The image images also leads shows to differences significant differences in imagesharpness, in brightness, which highlighted is indicated by by circle arrow in Figure in Figure 8, which 8. The are image not immanent also shows to significant object differences itself, but in due brightness, to bad illumination. highlighted by Again, circle inhuman Figure eye 8, which recognizes are not immanent se effects to as object interferences itself, but and dueis tostill bad able illumination. to deduce Again, underlying human eye information. recognizes se In an effects automated as interferences process, and however, is stillse able toeffects deduce interferences underlyingneed information. to be explicitly In an automated addressed. process, Consequently, however, se for reliable effects image interferences capturing, need it must to be explicitly be adjusted addressed. to fit Consequently, analytical process. for reliable This image also includes capturing, it must parallel be adjusted alignment to fit pictorial analytical object process. to This vertical also includes image edge. parallel alignment pictorial object to vertical Noticing image need edge. expert knowledge for identification relevant patterns, as well as for elimination Noticing need interferences, expert knowledge utilization for a identification knowledge base relevant was inevitable patterns, for as well species as for determination elimination in this project. interferences, utilization a knowledge base was inevitable for species determination Table 1 summarizes in this project. postulated assessment starting conditions. Table 1 summarizes postulated assessment starting conditions. Table 1. Pro and contra for automatised image processing application. Table 1. Pro and contra for automatised image processing application. Aspects Favouring Automatized Image Processing Challenges Aspects - defined Favouring illumination Automatized conditions Image Processing - Challenges different object sizes - simple patterns - differences in alignment - defined illumination conditions - different object sizes easy acquisition objects for knowledge base - differences due to biological variability - simple patterns - differences in alignment - easy acquisition objects for knowledge base - differences due to biological variability It appears worthwhile to enhance task species determination using methods image processing. From applications point view, task is to assign a three-dimensional object,

7 Insects 2017, 8, It appears worthwhile to enhance task species determination using methods image processing. applications point view, task is to assign a three-dimensional InsectsFrom object, Insects 2017, 8,2017, 52 8, larva, to its class hierarchy, consisting family, genus, and species. Biological variability and larva, to its class hierarchy, consisting family, genus, and species. Biological variability and larva, to its class hierarchy, consisting taken family,into genus, and species. Biological interferences image processing need to be consideration (Figure 9). variability and interferences image processing need to be taken into consideration (Figure 9). interferences image processing need to be taken into consideration (Figure 9). Insects 2017, 8, larva, to its class hierarchy, consisting family, genus, and species. Biological variability and interferences image processing need to be taken into consideration (Figure 9). Figure 9. Specification problem definition and primary influential factors. Figure 9. Specification problem definition and primary influential factors. Figure 9. Specification 4. The Image Processing Stware problem definition and primary influential factors. 4. The ImageBuilding Processing on Stware generally applicable hierarchy image processing operations [53], steps 4. The Image Processing Stware shown in Figure 10 are necessary to solve task using methods image processing. 9. Specification problem definition primary influential factors. Building on Figure generally applicable hierarchy and image processing operations [53], steps Building on generally applicable hierarchy image processing operations [53], steps shown in Figure areare necessary methodsimage image processing. shown in Figure necessarytotosolve solve task task using using methods processing. 4. The Image Processing Stware Building on generally applicable hierarchy image processing operations [53], steps shown in Figure 10 are necessary to solve task using methods image processing. Figure 10. Solution from an image processing perspective Filtration The aim filtration processes (Figure 11) is to emphasize patterns which are to be used for classification and to Figure remove10. any image perturbations. In or words, interfering signals should be Solution from an an image Figure 10. Solution from imageprocessing processingperspective. perspective. suppressed. Starting with original image (a), three steps are necessary to achieve this goal: 10.filtration Solution from an image processing perspective. high-pass filtration (b),figure median as preparation for generation a binary image (c) Filtration 4.1. Filtration 4.1. Filtration The aim filtration processes (Figure 11) is to emphasize patterns which are to be used for The aim and filtration processes (Figure 11) is to emphasize patterns which are to bebeused classification to remove any image perturbations. In or words, interfering signals should The aim filtration processes (Figure 11) is to emphasize patterns which are to be used for for classification and to remove any image perturbations. In or words, interfering signals should suppressed. Starting any original (a), three stepswords, are necessary tosignals achieve this be goal: classification and to with remove image image perturbations. In or interfering should be suppressed. Starting filtration originalas image three steps necessary to achieve this goal: high-pass filtration (b),with median preparation for generation atobinary image suppressed. Starting with original image (a),(a), three steps are are necessary achieve this (c). goal: high-pass filtration (b), median filtration for generation generation a binary image high-pass filtration (b), median filtrationasaspreparation preparation for a binary image (c). (c). Figure 11. Steps filtration process: Original image (a), high-pass filtration (b) and binary image (c). Figure 11. Steps filtration process: Original image (a), high-pass filtration (b) and binary image (c). Figure 11. Steps filtration process: Original image (a), high-pass filtration (b) and binary image (c). Figure 11. Steps filtration process: Original image (a); high-pass filtration (b); and binary image (c).

8 Insects 2017, 8, High-Pass Filtration to Emphasize Pattern The prerequisite for automated identification is a good perceptibility patterns against background. Good visibility is necessary to obtain characteristics desired signal. The light/dark boundary image allows for separation frequencies. In image processing, se boundaries are called edges. The image undergoing filtration process is considered to be a two-dimensional signal. The edges correspond to high frequencies, which need to be emphasized. In signal processing, this can be achieved using high-pass filtration. A Sobel filter was chosen for present application. This differentiation filter 1st order implements an approximation first derivative, and smoothing image in orthogonal direction to derivative. The filter operation consists 3 3 matrices for x and y direction. S x = , S y = The filtration involves convolution image matrix A with operators S x and S y, and results in filtered image composed sum individual convolution results. A xy = Median Filtration to Reduce Image Noise (S x A) 2 + ( S y A ) 2 The high-pass filtration emphasized desired signal. Since noise in image also consists high frequency sections, y, too, are now emphasized. They must refore be suppressed separately using a median filter. The median for each pixel with its adjacent neighbor is calculated, resulting in nine values. The median value in this list represents new brightness filtered pixel. At image margin, number neighbors is reduced to five, and at corner, to three Generating a Binary Image to Compare with Knowledge Base The knowledge base consists binary images. The patterns that are to be compared must refore be converted into same format. After application steps 1 and 2, image still includes 256 brightness levels. The application a static or dynamic threshold allows for conversion se brightness levels into binary values. A static threshold with a dynamic component was used in this case. The dynamic component is derived from average brightness all pixels. The static component is given by an fset value, which was set to an initial value 20. This is a proper value for well exposed and sharp images and was determined empirically Area Identification The subdivision individual patterns from filtered image is currently done semi-automatically. The user identifies relevant areas by manually drawing outline as free form. Each selection containing a single pattern is n automatically stored as an individual image and labelled accordingly. The resulting images are furr filtered for disruptive pixel groups, removing small groups pixels that could negatively influence classification. The last step process is automatic cropping and scaling pattern images to 100 pixels in height. This adjustment takes account size differences in larvae according to species and state development. Area identification generates up to nine images per segment. For each animal, images all segments analyzed are stored in one extensible-markup-language-file (XML-file), sorted according to location pattern in each segment. These files provide base for following classification tasks and enable visual verification image processing steps.

9 Insects 2017, 8, Classification via Knowledge Base The classification patterns requires generation and application a knowledge base. Sample images larvae belonging to 14 distinct species (Table 2) were selected and run through steps described in Section 4.1 (Filtration). To achieve a high level resolution, smallest possible meaningful elements, i.e., individual patterns (as shown in Figure 12 and previously explained in Figure 8), were chosen to construct knowledge base. Table 2. Results species determination. The results are listed alphabetically and given as a ratio correctly identified larvae to total number larvae analyzed. The highlighted fields indicate trials Insects 2017, 8, in which not all larvae were identified correctly. Trials marked with a dash (/) could not be executed 4.3. Classification via Knowledge Base due to lack suitable images required quality. The classification patterns requires generation and application a knowledge base. Sample images larvae belonging to 14 distinct species (Table 2) were selected and run through steps described in Section 4.1 (Filtration). To achieve a high level resolution, smallest possible meaningful elements, i.e., individual patterns (as shown in Figure 12 and previously explained No. Species Group A Group B in Figure 8), were chosen to construct knowledge Trial base. Result Trial Result 01 Calliphora vicina A1 14/15 B1 14/15 02 Table Lucilia 2. Results ampullacea species determination. The / results are listed alphabetically / and given B2 as a ratio 4/7 correctly identified larvae to total number larvae analyzed. The highlighted fields indicate 03 Lucilia illustris / / B3 6/8 trials in which not all larvae were identified correctly. Trials marked with a dash (/) could not be 04 executed Lucilia due to richardsi lack suitable images / required quality. / B4 7/7 05 Lucilia silvarum / / B5 4/4 No. Species Group A Group B 06 Sarcophaga albiceps / Trial Result / Trial B6 Result 10/ SarcophagaCalliphora argyrostoma vicina / A1 14/15 / B1 B7 14/15 7/7 02 Lucilia ampullacea / / B2 4/7 08 Sarcophaga caerulescens / B8 9/10 03 Lucilia illustris / / B3 6/ Sarcophaga Lucilia similis richardsi / / / B4 B9 7/7 5/ Wohlfarthia Lucilia silvarum bella A10 / 3/3 / B5 B10 4/4 7/7 06 Sarcophaga albiceps / / B6 10/10 11 Wohlfarthia indigens A11 2/2 B11 6/6 07 Sarcophaga argyrostoma / / B7 7/ Wohlfarthia Sarcophaga nuba caerulescens / / / B8 B12 9/10 3/ Wohlfarthia Sarcophaga trina similis A13 / 2/2 / B9 B13 5/6 9/10 10 Wohlfarthia bella A10 3/3 B10 7/7 14 Wohlfarthia villeneuvi / / B14 12/12 11 Wohlfarthia indigens A11 2/2 B11 6/6 12 Wohlfarthia nuba / / B12 3/3 13 Wohlfarthia trina A13 2/2 B13 9/10 14 Wohlfarthia villeneuvi / / B14 12/12 Reference patterns for each species are generated by user controlled iterative layering resulting sample images Reference (Figure patterns 12). for each species are generated by user controlled iterative layering resulting sample images (Figure 12). Figure 12. Layering several individual patterns (a) to generate a reference pattern (b). Figure 12. Layering several individual patterns (a) to generate a reference pattern (b). The user is able to remove distorting images or to manually correct images. Such interventions are needed, especially when patterns are heavily twisted or have an unusual size. The construction reference patterns is refore strongly influenced by experience and knowledge person conducting process. Consequently, knowledge base represents expert know-how. In analogy to analysis images, reference images need to be cropped and scaled to 100 pixels in height and saved as XML-files. These files constitute base for computational The user is able to remove distorting images or to manually correct images. Such interventions are needed, especially when patterns are heavily twisted or have an unusual size. The construction reference patterns is refore strongly influenced by experience and knowledge person implementation knowledge base. conducting process. Consequently, knowledge base represents expert know-how. In analogy to analysis images, reference images need to be cropped and scaled to 100 pixels in height and saved as XML-files. These files constitute base for computational implementation knowledge base. To ultimately be able to perform species determination in images from unknown larvae, classification method needs to be detailed. The calculation a normative cross-correlation

10 Insects 2017, 8, Insects 2017, To 8, 52 ultimately be able to perform species determination in images from unknown larvae, classification method needs to be detailed. The calculation a normative cross-correlation coefficient [54] between patterns binary image Axy and reference object Hxy is basis coefficient classification [54] betweenused patterns here. binary image A xy and reference object H xy is basis classification used here., =,, +, + k(x, y) = j,i H, xy (i, j)a +, + xy (x + i, y + j) kmax is determined for each pattern, resulting H xy (i, j) 2 in a characteristic A xy (x + i, y + j) 2 gradient for each reference image (Figure 13). k max An is determined average is calculated for each pattern, based on resulting this gradient in a characteristic and used for evaluation. gradient for The each reference image image (Figure with 13). highest average correlation to analysis image represents fly species that larva in question supposedly belongs to. Figure 13. Graphs correlation for two patterns. Figure 13. Graphs correlation for two patterns. 5. AnResults average is calculated based on this gradient and used for evaluation. The reference image with highest The original average images correlation larvae from to 14 analysis known species image represents were sorted into fly three species groups that for larva test in question run supposedly stware. belongs Group to. A consisted high quality images containing a full set patterns. Group B contained acceptable images missing single patterns. Group C contained images poor 5. Results quality in which a large share patterns could not be extracted for reasons poor lighting, insufficient The original contrast, imagesor discolorations. larvae from 14 known species were sorted into three groups for test run stware. The images Group from A group consisted A and B high were quality used to compile images references containing for a each full set species, patterns. where furr Group B differentiations were made: Class A references were exclusively composed group A images. If contained acceptable images missing single patterns. Group C contained images poor quality in number A images was insufficient for generation reliable reference patterns, class B which a large share patterns could not be extracted for reasons poor lighting, insufficient references were composed eir group B images or a mix A and B images. If necessary, C contrast, or discolorations. references were composed B and C images. The While images generating from group Aknowledge and B were base, used single to compile images had references to be removed for eachfrom species, layering where due furr differentiations to significant were contortion. made: Class Two species A references (Sacrophaga weremelanura exclusively and composed Lucilia sericata) group had to Abe images. excluded If number altoger A images due to was lack insufficient suitable for images. generation Patterns were reliable visible, reference but could patterns, only be class extracted B references in werepoor composed quality eir or not at all. group For Bsome images species, or agroup mix A Aand andb Bcould images. be employed, If necessary, and Cfor references ors, only were composed group B could B andbe Cused. images. The test compared each image to each reference. While generating knowledge base, single images had to be removed from layering due to significant 5.1. Results contortion. Species Determination Two species (Sacrophaga melanura and Lucilia sericata) had to be excluded altoger The duethree to references lack suitable with images. highest Patterns average were correlation visible, coefficients, but could only as well be extracted as three in poor quality references or not atwith all. For some highest species, sum groups correlation A and B could coefficients, be employed, were and selected for ors, as only results. group B coulda be total used. The 134 larvae test compared from 14 each different image species to each were reference. analyzed, which 124 (92.5%) could be identified correctly (Table 2) Results Species Determination The three references with highest average correlation coefficients, as well as three references with highest sum correlation coefficients, were selected as results. A total 134 larvae from 14 different species were analyzed, which 124 (92.5%) could be identified correctly (Table 2).

11 Insects 2017, 8, Results Genus Determination The three references with highest correlation coefficient are furr investigated. The result is conceded as true if three references belong to same genus; orwise, no genus could be determined. The test included 97 larvae from 13 different species, where 68 larvae (70%) were identified correctly (Table 3). Table 3. Results genus determination. The results are listed alphabetically and given as a ratio correctly identified larvae to total number larvae analyzed. The highlighted fields indicate trials in which not all larvae were identified correctly. Trials marked with a dash (/) could not be executed due to lack suitable images required quality. No. Species Group A Group B Trial Result Trial Result 01 Lucilia ampullacea / / B2 6/7 02 Lucilia illustris / / B3 8/8 03 Lucilia richardsi / / B4 3/7 04 Lucilia silvarum / / B5 3/4 05 Sarcophaga albiceps / / B6 6/10 06 Sarcophaga argyrostoma / / B7 1/7 07 Sarcophaga caerulescens / / B8 0/10 08 Sarcophaga similis / / B9 3/6 09 Wohlfarthia bella A10 3/3 B10 7/7 10 Wohlfarthia indigens A11 2/2 B11 6/6 11 Wohlfarthia nuba / / B12 3/3 12 Wohlfarthia trina A13 2/2 B13 10/10 13 Wohlfarthia villeneuvi / / B14 12/12 One requirement is mandatory for genus determination: The references same genus must be more similar to one anor than to those any or genus. To validate this requirement, all references were compared to one anor. For each reference, three references with highest correlation coefficients are given (extract in Table 4). Each genus was represented by at least four species. For a correct determination, all three correlating reference species must refore be members same genus. Table 4. Similarities between references (extract). The results are listed alphabetically; numbers given are correlation coefficients compared to reference. Highlighted fields indicate trials in which genus adjacent references differed from genus reference analyzed. Reference Lucilia ampullacea Lucilia illustris Lucilia richardsi Lucilia silvarum Sarcophaga albiceps Sarcophaga argyrostoma Sarcophaga caerulescens Sarcophaga similis Wohlfarthia bella Wohlfarthia indigens Wohlfarthia nuba Wohlfarthia trina Wohlfarthia villeneuvi References Adjacent Similarity Lucilia richardsi Lucilia silvarum Lucilia illustris (0.7944) (0.7843) (0.7759) Lucilia richardsi Lucilia silvarum Lucilia ampullacea (0.7994) (0.7882) (0.7872) Lucilia ampullacea Lucilia illustris Lucilia silvarum (0.8040) (0.7930) (0.7919) Lucilia richardsi Lucilia ampullacea Lucilia illustris (0.7950) (0.7890) (0.7696) Sarcophaga similis Wohlfarthia indigens Sarcophaga caerulescens (0.7137) (0.7070) (0.7061) Sarcophaga caerulescens Sarcophaga similis Lucilia richardsi (0.7366) (0.7305) (0.7243) Sarcophaga similis Wohlfarthia trina Wohlfarthia villeneuvi (0.7612) (0.7598) (0.7489) Sarcophaga caerulescens Wohlfarthia indigens Lucilia richardsi (0.7714) (0.7526) (0.7472) Wohlfarthia villeneuvi Wohlfarthia trina Wohlfarthia indigens (0.7671) (0.7566) (0.7382) Wohlfarthia bella Wohlfarthia villeneuvi Wohlfarthia trina (0.7402) (0.7343) (0.7341) Wohlfarthia indigens Wohlfarthia trina Wohlfarthia villeneuvi (0.7725) (0.7492) (0.7402) Wohlfarthia indigens Wohlfarthia villeneuvi Wohlfarthia nuba (0.7833) (0.7758) (0.7641) Wohlfarthia indigens Wohlfarthia trina Wohlfarthia bella (0.7976) (0.7940) (0.7792)

12 Insects 2017, 8, The highlighted fields indicate samples in which this could not be achieved. This suggests that differing genus references might display more similarities to investigated sample than references corresponding genus. The identification genera Lucilia and Sacrophaga proved to be difficult. The genus Wohlfarthia, on or hand, could always be determined correctly (Table 4) Problem Analysis All following problems (Table 5) were identified during process. The problems do not generally occur and combinations are possible. Table 5. Recognized problems and possible solutions. Problem The patterns contain many holes. In holey patterns parts border may be missing. This affects identification. Additionally, holey patterns may mislead interpretation layered image. The result is a reference pattern with poor properties. The patterns reference are too voluminous. This changes correlation coefficient. The pattern to be analysed is too voluminous. In this case pattern ten has holes. This changes correlation coefficient. Misinterpretation central patterns. The central pattern consists two or more partial patterns. Those vary significantly in size and distance within one species. In addition, patterns species same genus show many similarities. Solution The combination high-pass filter and threshold should be adjusted. The aim is to recognise each pattern point as closed, hollow form. Afterwards a dilatation algorithm is applied. Guidelines for generation references need to be set. Firstly, no pattern with holes should be used. Furrmore, it is important to layer at least 10 patterns, to generate a valid reference. Additionally, a new tool should be developed to allow rotation patterns. A predefined area can be built around geometric centre pattern point. This new pattern can be used instead pattern point. Three possible solutions could be investigated: Firstly, it should be tested wher central pattern can be ignored. If this is not possible it should be scaled in width. If this also does not yield any favourable results, it might be tested wher central pattern can be split along axis symmetry. 6. Discussion and Outlook The aim our work was to improve method MAS pattern recognition for species and age identification in larval instars Diptera. The MAS method is based on comparison an unknown pattern ( a larva in question) with existing reference patterns (generated from patterns a number known larvae). A broad data base reference patterns is needed for this task. Facilitation MAS pattern recognition allows for much faster generation reference patterns and thus, expansion knowledge base. At first glance, use or methods than a knowledge base sounds promising. Complex identification problems can ten be solved with methods so called artificial intelligence. This approach, however, had to be declined for following reasons. First, it is hardly possible to create a suitable amount learning data to meet claim, e.g., a neural network. This is due to intricate process manual preparation larvae. We are talking about at least 100 perfect samples to support learning process. In probabilistic case failure or a reduced accuracy learning process, this number would have to be increased. The second reason is basic necessity a coefficient determination. The proposed method provides average correlation coefficient, while neural networks cannot evaluate accuracy results image processing by mselves. Entomological scientists from around world might be needed to contribute and analyze a vast pool known larvae for MAS method. To facilitate this, protocols need to be established

13 Insects 2017, 8, for standardization in preparation larvae, production original pictures, and use stware. All hirto existing preparations were photographed with a digital camera mounted onto dissecting microscope. Special care was given to vertical adjustment ventral midline for a better comparison patterns. Lighting aspects, however, were no particular importance while analyzing pictures by hand and in small numbers. This resulted in scattered illumination on original pictures. To avoid editing original pictures prior to automated analysis, a new protocol has to be developed. Pictures need to be recorded in optimal lighting conditions with combined reflected and transmitted illumination. This is possible when cuticula is pinned to a clear base, like two-component Sylgard silicone, using very fine tungsten pins. Optimal orientation must be carried out even more accurately. This can be achieved by use alignment instruments like cross hair eyepieces. Caution is recommended when increasing staining time to reach more high-contrast pictures, as surrounding areas or whole cuticula might be discolored. An additional feature allowing upload manually charted MAS patterns would furr increase usability database. Standardized patterns might considerably improve genus determination, as influence small structures would be reduced. Furrmore, MAS patterns in specimens with discolorations or or disturbances could be included. In particular this concerns pictures which were classified as C images due to disadvantageous features found in larva, not because image capturing problems. Such disturbances might interfere with computerized analysis. A number more traditional morphological species indicators are located on cuticula larval Diptera, such as spinal bands and spiracles. While preparing cuticula according to protocol, such indicators could automatically be included. If due to close relatedness or or factors results MAS analysis are not unambiguous, those or morphological characteristics might be used to substantiate species or genus identification. The MAS method thus complements traditional methods, rar than rivalling m. MAS methods, course, are not restricted to forensically important Diptera. Many over 200 known dipteran families might not be forensically important, but could be included in a MAS database for species determination in or fields entomology. Furrmore, if data base can be expanded, identification might not remain only objective. Not all dipteran larvae live in same habitat and have same behavior. Muscle equipment larvae might reflect such differences. An indication for furr analysis muscle equipment and behavior could be given by muscle attachment site pattern comparisons in different species. Obligatory parasitic larvae for instance might have a reduced number muscles due to diminished motility. Confirmation this statement, however, needs furr studies larvae representing different breeding behaviors. Such a general view could lead to new insight in ecological systems. 7. Conclusions With growth knowledge base and a complete description all segments, re might not only arise more differences between species, but also some so far unknown similarities. In an ambitious project, a complete map could be drawn all clusters in all accessible developmental stages larvae. The tempting prospect would be that even a fragment a larva would be sufficient to determine its species and developmental state. An increase in precision stware will refore be essential, which in turn could lead to ability to distinguish seemingly indistinguishable species (e.g., Piophilinae species). Finally, with a broad knowledge base and corresponding stware, species determination in dipteran larvae would become a quick and performable task for every interested individual able to take a digital picture larval cuticula. No entomological training and/or piles field guides and keys with endless details would be needed!

14 Insects 2017, 8, Author Contributions: Senta Niederegger developed MAS method; Klaus-Peter Döge developed methodology image processing and test plan; Marcus Peter and Tobias Eickhölter performed experiments and programming; Gita Mall provided research materials and facilities; Senta Niederegger and Klaus-Peter Döge wrote paper. Conflicts Interest: The authors declare no conflict interest. References 1. Byrd, J.H.; Castner, J.L. Forensic Entomology: The Utility Arthropods in Legal Investigations, 2nd ed.; CRC Press: Boca Raton, FL, USA, 2009; p. 681, ISBN Greenberg, B.; Kunich, J.C. Entomology and Law. Flies as Forensic Indicators, Paperback, First ed.; Cambridge University Press: Cambridge, UK, 2002; 306p, ISBN Gennard, D.E. Forensic Entomology: An Introduction; Wiley: Chichester, UK, 2007; 223p, ISBN Lane, R.P. Investigation into Blowfly (Diptera-Calliphoridae) Succession on Corpses. J. Nat. Hist. 1975, 9, Catts, E.P.; Gf, M.L. Forensic entomology in criminal investigations. Annu. Rev. Entomol. 1992, 37, Carter, D.O.; Yellowlees, D.; Tibbett, M. Cadaver decomposition in terrestrial ecosystems. Naturwissenschaften 2007, 94, Reiter, C. Zum Wachstumsverhalten der Maden der blauen Schmeißfliege Calliphora vicina. Z. Rechtsmed. 1984, 91, Anderson, G.S. Minimum and maximum development rates some forensically important Calliphoridae (Diptera). J. Forensic Sci. 2000, 45, Kaneshrajah, G.; Turner, B. Calliphora vicina larvae grow at different rates on different body tissues. Int. J. Leg. Med. 2004, 118, Donovan, S.E.; Hall, M.J.R.; Turnerand, B.; Moncrieff, C.B. Larval Growth Rates Blowfly, Calliphora vicina, Over a Range Temperatures. Med. Vet. Entomol. 2006, 20, Niederegger, S.; Pastuschek, J.; Mall, G. Preliminary studies influence fluctuating temperatures on development various forensically relevant flies. Forensic Sci. Int. 2010, 199, Amendt, J.; Krettek, R.; Niess, C.; Zehnerand, R.; Bratzke, H. Forensic entomology in Germany. Forensic Sci. Int. 2000, 113, O Flynn, M.A.; Moorhouse, D.E. Identification early immature stages some common Queensland carrion flies. J. Aust. Entomol. Soc. 1980, 19, Cantrell, B.K. The Immature Stages some Australian Sarcophaginae (Diptera: Sarcophagidae). J. Aust. Entomol. Soc. 1981, 20, Reiter, C.; Wollenek, G. Zur Artbestimmung der Maden forensisch bedeutsamer Schmeißfliegen. Z. Rechtsmed. 1983, 90, Erzinçlioğlu, Y.Z. Immature stages British Calliphora and Cynomya, with a re-evaluation taxonomic characters larval Calliphoridae (Diptera). J. Nat. Hist. 1985, 19, Queiroz, M.M.D.; demello, R.P.; Lima, M.M. Morphological aspects larval instars Chrysomya albiceps (Diptera, Calliphoridae) reared in laboratory. Mem. Inst. Oswaldo Cruz 1997, 92, Wallman, J.F. Third-Instar Larvae Common Carrion-Breeding Blowflies Genus Calliphora (Diptera: Calliphoridae) in South Australia. Invertebr. Taxon. 2001, 15, Sukontason, K.L.; Piangjai, S.; Boonsriwong, W.; Bunchu, N.; Ngern-klun, R.; Vogtsbergerand, R.C.; Sukontason, K. Observations third instar larva and puparium Chrysomya bezziana (Diptera: Calliphoridae). Parasitol. Res. 2006, 99, Szpila, K.; Pape, T.; Rusinek, A. Morphology first instar Calliphora vicina, Phormia regina and Lucilia illustris (Diptera, Calliphoridae). Med. Vet. Entomol. 2008, 22, Martin-Vega, D.; Gomez-Gomez, A.; Baz, A.; Diaz-Aranda, L.M. New piophilid in town: first Palaearctic record Piophila megastigmata and its coexistence with Piophila casei in central Spain. Med. Vet. Entomol 2011, 25, Singh, D.; Garg, R.; Wadhawan, B. Ultramorphological characteristics immature stages a forensically important fly Parasarcophaga ruficornis (Fabricius) (Diptera: Sarcophagidae). Parasitol. Res. 2012, 110,

15 Insects 2017, 8, Wells, J.D.; Sperling, F.A.H. DNA-based identification forensically important Chrysomyinae (Diptera: Calliphoridae). Forensic Sci. Int. 2001, 120, Wells, J.D.; Pape, T.; Sperling, F.A.H. DNA-Based Identification and Molecular Systematics Forensically Important Sarcophagidae (Diptera). J. Forensic Sci. 2001, 46, Boehme, P.; Amendt, J.; Disney, R.H.L.; Zehner, R. Molecular identification carrion-breeding scuttle flies (Diptera: Phoridae) using COI barcodes. Int. J. Leg. Med. 2010, 124, Sukontason, K.L.; Piangjai, S.; Bunchu, N.; Chaiwong, T.; Sripakdee, D.; Boonsriwong, W.; Vogtsberger, R.C.; Sukontason, K. Surface ultrastructure puparia blow fly, Lucilia cuprina (Diptera: Calliphoridae), and flesh fly, Liosarcophaga dux (Diptera: Sarcophagidae). Parasitol. Res. 2006, 98, Anderson, G.S. Determining time death using blow fly eggs in early postmortem interval. Int. J. Leg. Med. 2004, 118, Crossley, A.C. Transformations in Abdominal Muscles Blue Blow-Fly Calliphora Erythrocephala (Meig) during Metamorphosis. J. Embryol. Exp. Morphol. 1965, 14, Hooper, J.E. Homeotic Gene-Function in Muscles Drosophila Larvae. EMBO J. 1986, 5, , PMCID:PMC Roberts, M.J. Locomotion Cyclorrhaphan Maggots (Diptera). J. Nat. Hist. 1971, 5, Hanslik, U.; Schos, A.; Niederegger, S.; Heinzel, H.G.; Spiess, R. The Thoracic Muscular System and Its Innervation in Third Instar Calliphora vicina Larvae. I. Muscles Pro- and Mesothorax and Pharyngeal Complex. J. Morphol. 2010, 271, Niederegger, S.; Spieß, R. Cuticular muscle attachment sites as a tool for species determination in blowfly larvae. Parasitol. Res. 2012, 110, Niederegger, S.; Miroschnikow, A.; Spieß, R. Marked for life: muscle attachment site patterns in blowfly larvae are constant throughout development. Parasitol. Res. 2013, 112, Niederegger, S.; Spieß, R. Muscle attachment sites Phormia regina (Meigen). Parasitol. Res. 2014, 113, Martin-Vega, D.; Baz, A.; Diaz-Aranda, L.M. The immature stages necrophagous fly, Prochyliza nigrimana: comparison with Piophila casei and medicolegal considerations (Diptera: Piophilidae). Parasitol. Res. 2012, 111, Sukontason, K.; Sribanditmongkol, P.; Ngoen-klan, R.; Klong-klaew, T.; Moophayak, K.; Sukontason, K.L. Differentiation between Lucilia cuprina and Hemipyrellia ligurriens (Diptera: Calliphoridae) larvae for use in forensic entomology applications. Parasitol. Res. 2010, 106, Niederegger, S.; Szpila, K.; Mall, G. Muscle attachment site (MAS) patterns for species determination in European species Lucilia (Diptera: Calliphoridae). Parasitol. Res. 2015, 114, Smith, K.G.V. A Manual Forensic Entomology; The Trustees British Museum (Natural History): London, UK, 1986; Volume Szpila, K. Key for identification third instars European blowflies (Diptera: Calliphoridae) forensic importance. In Current Concepts in Forensic Entomology; Amendt, J., Gf, M.L., Campobasso, C.P., Grassberger, M., Eds.; Springer: Dordrecht Heidelberg, Germany; London, UK; New York, NY, USA, 2010; pp , ISBN Szpila, K.; Hall, M.J.R.; Pape, T.; Grzywacz, A. Morphology and identification first instars European and Mediterranean blowflies forensic importance. Part II. Luciliinae. Med. Vet. Entomol. 2013, 27, Schumann, H. Die Gattung Lucilia (Goldfliegen). Merkbl. Angew. Parasitenkd. Jena 1971, 18, Sonet, G.; Jordaens, K.; Braet, Y.; Desmyter, S. Why is molecular identification forensically important blowfly species Lucilia caesar and L. illustris (family Calliphoridae) so problematic? Forensic Sci. Int. 2012, 223, Szpila, K.; Richet, R.; Pape, T. Third instar larvae flesh flies (Diptera: Sarcophagidae) forensic importance Critical review characters and key for European species. Parasitol. Res. 2015, 144, Anton, E.; Niederegger, S.; Beutel, R.G. Beetles and flies collected on pig carrion in an experimental setting in Thuringia and ir forensic implications. Med. Vet. Entomol. 2011, 25, Cherix, D.; Wyss, C.; Pape, T. Occurrences flesh flies (Diptera: Sarcophagidae) on human cadavers in Switzerland, and ir importance as forensic indicators. Forensic Sci. Int. 2012, 220, Grassberger, M.; Frank, C. Initial Study Arthropod Succession on Pig Carrion in a Central European Urban Habitat. J. Med. Entomol. 2004, 41,

16 Insects 2017, 8, Madra, A.; Fratczak, K.; Grzywacz, A.; Matuszewski, S. Long-term study pig carrion entomauna. Forensic Sci. Int. 2015, 252, Matuszewski, S.; Szafalowicz, M.; Jarmusz, M. Insects colonising carcasses in open and forest habitats Central Europe: Search for indicators corpse relocation. Forensic Sci. Int. 2013, 231, Pohjoismäki, J.L.; Karhunen, P.J.; Goebeler, S.; Saukko, P.; Saaksjarvi, I.E. Indoors forensic entomology: Colonization human remains in closed environments by specific species sarcosaprophagous flies. Forensic Sci. Int. 2010, 199, Szpila, K.; Madra, A.; Jarmusz, M.; Matuszewski, S. Flesh flies (Diptera: Sarcophagidae) colonising large carcasses in Central Europe. Parasitol. Res. 2015, 144, Niederegger, S.; Szpila, K.; Mall, G. Muscle attachment site (MAS) patterns for species determination in five species Sarcophaga (Diptera: Sarcophagidae). Parasitol. Res. 2016, 115, Martin-Vega, D.; Niederegger, S. Larval muscle attachment site (MAS) patterns are a conserved character among Piophilini flies (Diptera, Piophilidae). Dtsch. Entomol. Z. 2015, 62, Jähne, B. Digitale Bildverarbeitung; Springer: Berlin/Heidelberg, Germany; New York, NY, USA, 2002; pp , ISBN Maschotta, R. Merkmalslistenbasierte Kreuzkorrelationsmethoden für die medizinische Bildverarbeitung; Südwestdeutscher Verlag für Hochschulschriften: Saarbrücken, Germany, 2009; ISBN by authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under terms and conditions Creative Commons Attribution (CC BY) license (