Sex determination using the clavicle by deep learning in a Thai population.

Determining sex is a critical process in estimating biological profiles from skeletal remains. The clavicle is interesting in studying sex determination because it is durable to the environment, slow to decay, challenging to destroy, making the clavicle useful in autopsies and identification which can then lead to verification. The goal of this study was to use deep learning in determining sex from clavicles within the Thai population and obtain the accuracies for the validation set using a convolutional neural network (GoogLeNet). A total of 200 pairs of clavicles were obtained from 200 Thai persons (100 males and 100 females) as part of a training group. For the deep learning approach, the clavicle was photographed, and each clavicle image was submitted to the training model for sex determination. Training groups of 200 samples were made. Images of the same size were input into the training model. The percentage of the validation set accuracy was calculated from the MATLAB program. GoogLeNet was the best training model and get the result of validation set accuracy. The results of this study found accuracies for a validation set with the highest overall right lateral view of the clavicle with an accuracy of 95%. Accuracy from the validation set of each view of the clavicle can demonstrate the forensic value of sex determination. A deep learning approach with clavicles can determine the sex and is simple to utilize for forensic anthropology professionals.

Estimation of Stature Based on Metatarsal Bones in a Thai Population / Estimación de la Estatura Basada en los Huesos Metatarsianos en una Población Tailandesa

SUMMARY: Stature estimation is one of the essential procedures for personal identification in forensic osteology. Therefore, the purposes of this study are to analyze the correlation between length and width of metatarsal measurements and stature, and to develop the regression equations for a Thai population. In this study, the samples were divided into two groups. The first group was called the "training group" for generating stature estimation equations, comprised of 200 skeletons, aged between 19-94 years. The second group was called the "test group" for evaluating the accuracy of generated equations, comprising 40 skeletons. The correlation between metatarsal parameters and stature were moderate to high, and all variables had positive significant correlation with stature. For males, the left ML2 is the length variable that showed the most correlation degree against stature (r=0.702), and the left MSW4 is the width variable that had the most correlation degree against stature (r=0.483). For females, right ML1 is the length variable that had the most correlation degree against stature (r=0.632), and right PW3 is the width stature that had the most correlation degree against stature (r=0.481). For all samples, left ML1 was the length variable that had the most correlation degree against stature (r=0.796) and right PW3 was the width variable that had the most correlation degree against stature (r=0.712). The results of generating multiple regression equations using a stepwise method reveals that the correlation coefficient (R) and standard error of estimate (SEE) were 0.761 and 4.96 cm, respectively, for males, and 0.752 and 4.93 cm for females, with 0.841 and 5.26 cm for all samples, respectively. According to these results, the mean of absolute error from the test group ranged from 3 to 5 cm. Therefore, stature estimation equations using length and width of metatarsals from our study can be applied to estimate stature in the Thai population.

La estimación de la estatura es uno de los procedimientos esenciales para la identificación personal en osteología forense. Por lo tanto, los propósitos de este estudio fueron analizar la correlación entre la longitud y el ancho de las medidas metatarsianas y la estatura, y desarrollar las ecuaciones de regresión para una población tailandesa. Las muestras se dividieron en dos grupos. El primer grupo se denominó "grupo de entrenamiento" para generar ecuaciones de estimación de estatura, compuesto por 200 esqueletos, con edades comprendidas entre los 19 y los 94 años. El segundo grupo se denominó "grupo de prueba" para evaluar la precisión de las ecuaciones generadas, que comprende 40 esqueletos. La correlación entre los parámetros metatarsianos y la estatura fue de moderada a alta, y todas las variables tuvieron una correlación significativa positiva con la estatura. Para el sexo masculino, la variable longitud ML2 izquierda es la que mayor grado de correlación presentó con la estatura (r=0,702), y la izquierda MSW4 fue la variable ancho la que mayor grado de correlación presentó con la estatura (r=0,483). Para el sexo femenino, ML1 derecho fue la variable longitud que tuvo mayor grado de correlación con la estatura (r=0,632), y PW3 derecha fue la variable ancho estatura que tuvo mayor grado de correlación con la estatura (r=0,481). Para todas las muestras, ML1 izquierdo fue la variable longitud que tuvo mayor grado de correlación con la estatura (r=0,796) y PW3 derecha fue la variable ancho que tuvo mayor grado de correlación con la estatura (r=0,712). Los resultados de generar ecuaciones de regresión múltiple usando un método paso a paso revela que el coeficiente de correlación (R) y el error estándar de estimación (SEE) fueron 0,761 y 4,96 cm, respectivamente, para los hombres y 0,752 y 4,93 cm para las mujeres, con 0,841 y 5,26 cm para todas las muestras, respectivamente. De acuerdo con estos resultados, la media del error absoluto del grupo de prueba osciló entre 3 y 5 cm. Por lo tanto, las ecuaciones de estimación de la estatura que utilizan la longitud y el ancho de los metatarsianos de nuestro estudio se pueden aplicar para estimar la estatura en la población tailandesa.

Cadaveric study investigating the femoral nerve-sparing volume for pericapsular nerve group (PENG) block.

BACKGROUND: This cadaveric study investigated the maximum effective volume of dye in 90% of cases (MEV90) required to stain the iliac bone between the anterior inferior iliac spine (AIIS) and the iliopubic eminence (IPE) while sparing the femoral nerve during the performance of pericapsular nerve group (PENG) block. METHODS: In cadaveric hemipelvis specimens, the ultrasound transducer was placed in a transverse orientation, medial and caudal to the anterior superior iliac spine in order to identify the AIIS, the IPE and the psoas tendon. Using an in-plane technique and a lateral-to-medial direction, the block needle was advanced until its tip contacted the iliac bone. The dye (0.1% methylene blue) was injected between the periosteum and psoas tendon. Successful femoral-sparing PENG block was defined as the non-staining of the femoral nerve on dissection. Volume assignment was carried out using a biased coin design, whereby the volume of dye administered to each cadaveric specimen depended on the response of the previous one. In case of failure (ie, stained femoral nerve), the next one received a lower volume (defined as the previous volume with a decrement of 2 mL). If the previous cadaveric specimen had a successful block (ie, non-stained femoral nerve), the next one was randomized to a higher volume (defined as the previous volume with an increment of 2 mL), with a probability of b=1/9, or the same volume, with a probability of 1-b=8/9. RESULTS: A total of 32 cadavers (54 cadaveric hemipelvis specimens) were included in the study. Using isotonic regression and bootstrap CI, the MEV90 for femoral-sparing PENG block was estimated to be 13.2 mL (95% CI: 12.0 to 20.0). The probability of a successful response was estimated to be 0.93 (95% CI: 0.81 to 1.00). CONCLUSION: For PENG block, the MEV90 of methylene blue required to spare the femoral nerve in a cadaveric model is 13.2 mL. Further studies are required to correlate this finding with the MEV90 of local anesthetic in live subjects.

Convolutional neural network of age-related trends digital radiographs of medial clavicle in a Thai population: a preliminary study.

Age at death estimation has always been a crucial yet challenging part of identification process in forensic field. The use of human skeletons have long been explored using the principle of macro and micro-architecture change in correlation with increasing age. The clavicle is recommended as the best candidate for accurate age estimation because of its accessibility, time to maturation and minimal effect from weight. Our study applies pre-trained convolutional neural network in order to achieve the most accurate and cost effective age estimation model using clavicular bone. The total of 988 clavicles of Thai population with known age and sex were radiographed using Kodak 9000 Extra-oral Imaging System. The radiographs then went through preprocessing protocol which include region of interest selection and quality assessment. Additional samples were generated using generative adversarial network. The total clavicular images used in this study were 3,999 which were then separated into training and test set, and the test set were subsequently categorized into 7 age groups. GoogLeNet was modified at two layers and fine tuned the parameters. The highest validation accuracy was 89.02% but the test set achieved only 30% accuracy. Our results show that the use of medial clavicular radiographs has a potential in the field of age at death estimation, thus, further study is recommended.

Deep learning and morphometric approach for Sex determination of the lumbar vertebrae in a Thai population.

Sex determination is a fundamental step in biological profile estimation from skeletal remains in forensic anthropology. This study proposes deep learning and morphometric technique to perform sex determination from lumbar vertebrae in a Thai population. A total of 1100 lumbar vertebrae (L1-L5) from 220 Thai individuals (110 males and 110 females) were obtained from the Forensic Osteology Research Center, Faculty of Medicine, Chiang Mai University, Thailand. In addition, two linear measurements of superior and inferior endplates from the digital caliper and image analysis were carried out for morphometric technique. Deep learning applied image classification to the superior and inferior endplates of the lumbar vertebral body. All lumbar vertebrae images are included in the dataset to increase the number of images per class. The accuracy determined the performance of each technique. The results showed the accuracies of 82.7%, 90.0%, and 92.5% for digital caliper, image analysis, and deep learning techniques, respectively. The lumbar vertebrae L1-L5 exhibit sexual dimorphism and can be used in sex estimation. Deep learning is more accurate in determining sex than the morphometric method. In addition, the subjectivity and errors in the measurement are decreased. Finally, this study presented an alternative approach to determining sex from lumbar vertebrae when the more traditionally used skeletal elements are incomplete or absent.

Comparison of sex determination using three methods applied to the greater sciatic notch of os coxae in a Thai population: Dry bone morphology, 2-dimensional photograph morphometry, and deep learning artificial neural network.

The os coxa is commonly used for sex and age estimation with a high degree of accuracy. Our study aimed to compare the accuracy among three methods, which include a deep learning approach to increase the accuracy of sex prediction. A total sample of 250 left os coxae from a Thai population was divided into a 'training' set of 200 samples and a 'test' set of 50 samples. The age of the samples ranged from 26 to 94 years. Three methods of sex determination were assessed in this experiment: a dry bone method, an image-based method and deep learning method. The intra- and inter-observer reliabilities were also assessed in the dry bone and photo methods. Our results showed that the accuracies were 80.65%, 90.3%, and 91.95% for the dry bone, image-based, and deep learning methods, respectively. The greater sciatic notch shape was wide and symmetrical in females while narrow and asymmetrical in males. The intra- and inter-observer agreements were moderate to almost perfect level (Kappa = 0.67-0.93, ICC = 0.74-0.94). Conclusion: The image-based and deep learning methods were efficient in sex determination. However, the deep learning technique performed the best among the three methods due to its high accuracy and rapid analysis. In this study, deep learning technology was found to be a viable option for remote consultations regarding sex determination in the Thai population.

A comparative study of visual assessment between dry bone, 2-dimensional photograph, and deep learning methods in sex classification on the auricular area of the os coxae in a Thai population / Un estudio comparativo de la evaluación visual entre el hueso seco, la fotografía bidimensional y los métodos de aprendizaje profundo en la clasificación del sexo en el área auricular del hueso coxal en una población tailandesa

SUMMARY: Sex assessment is an important process in forensic identification. A pelvis is the best skeletal element for identifying sexes due to its sexually dimorphic morphology. This study aimed to compare the accuracy of the visual assessment in dry bones as well as 2D images and to test the accuracy of using a deep convolutional neural network (GoogLeNet) for increasing the performance of a sex determination tool in a Thai population. The total samples consisted of 250 left os coxa that were divided into 200 as a 'training' group (100 females, 100 males) and 50 as a 'test' group. In this study, we observed the auricular area, both hands-on and photographically, for visual assessment and classified the images using GoogLeNet. The intra-inter observer reliabilities were tested for each visual assessment method. Additionally, the validation and test accuracies were 85, 72 percent and 79.5, 60 percent, for dry bone and 2D image methods, respectively. The intra- and inter-observer reliabilities showed moderate agreement (Kappa = 0.54 - 0.67) for both visual assessments. The deep convolutional neural network method showed high accuracy for both validation and test sets (93.33 percent and 88 percent, respectively). Deep learning performed better in classifying sexes from auricular area images than other visual assessment methods. This study suggests that deep learning has advantages in terms of sex classification in Thai samples.

RESUMEN: La evaluación del sexo es un proceso importante en la identificación forense. La pelvis es el mejor elemento esquelético para identificar sexos debido a su morfología sexualmente dimórfica. Este estudio tuvo como objetivo comparar la precisión de la evaluación visual en huesos secos, así como imágenes 2D y probar la precisión del uso de una red neuronal convolucional profunda (GoogLeNet) para aumentar el rendimiento de una herramienta de determinación de sexo en una población tailandesa. Las muestras consistieron en 250 huesos coxales izquierdos, los que fueron dividi- das de la siguiente manera: 200 como un grupo de "entrenamiento" (100 mujeres, 100 hombres) y 50 como un grupo de "prueba". En este estudio, observamos el área auricular, tanto de forma práctica como fotográfica, para una evaluación visual y clasificamos las imágenes utilizando GoogLeNet. Se analizó la confiabilidad intra-interobservador para cada método de evaluación visual. Además, las precisiones de validación y prueba fueron del 85, 72 por ciento y 79,5, 60 por ciento, para los métodos de hueso seco y de imágenes 2D, respectivamente. Las confiabilidades intra e interobservador mostraron un acuerdo moderado (Kappa = 0.54 - 0.67) para ambas evaluaciones visuales. El método de red neuronal convolucional profunda mostró una alta precisión tanto para la validación como para los conjuntos de prueba (93,33 por ciento y 88 por ciento, respectivamente). El aprendizaje se desempeñó mejor en la clasificación de sexos a partir de imágenes del área auricular que otros métodos de evaluación visual. Este estudio sugiere que el aprendizaje profundo tiene ventajas en términos de clasificación por sexo en muestras tailandesas.