Archives of academic emergency medicine · 2024
Detection of Body Packs in Abdominal CT scans Through Artificial Intelligence; Developing a Machine Learning-based Model.
СтатьяНазвание и аннотация: Перевод пока не подготовлен
Обработка статьи · этапы, ошибки и управление
Загрузка состояния…
Ошибки и ограничения
История · было → стало
Состояние обновляется каждые 5 секунд, пока блок открыт. Процент показывается только там, где известен объём работы. Пауза останавливает продолжение этапа; текущий запрос к источнику может завершиться.
Hosseini Sayed Masoud, Mohtarami Seyed Ali, Shadnia Shahin, Rahimi Mitra, Erfan Talab Evini Peyman, Mostafazadeh Babak, Memarian Azadeh, Heidarli Elmira
Аннотация
INTRODUCTION: Identifying the people who try to hide illegal substances in the body for smuggling is of considerable importance in forensic medicine and poisoning. This study aimed to develop a new diagnostic method using artificial intelligence to detect body packs in real-time Abdominal computed tomography (CT) scans.
METHODS: In this cross-sectional study, abdominal CT scan images were employed to create a machine learning-based model for detecting body packs. A single-step object detection called RetinaNet using a modified neck (Proposed Model) was performed to achieve the best results. Also, an angled Bbox (oriented bounding box) in the training dataset played an important role in improving the results.
RESULTS: A total of 888 abdominal CT scan images were studied. Our proposed Body Packs Detection (BPD) model achieved a mean average precision (mAP) value of 86.6% when the intersection over union (IoU) was 0.5, and a mAP value of 45.6% at different IoU thresholds (from 0.5 to 0.95 in steps of 0.05). It also obtained a Recall value of 58.5%, which was the best result among the standard object detection methods such as the standard RetinaNet.
CONCLUSION: This study employed a deep learning network to identify body packs in abdominal CT scans, highlighting the importance of incorporating object shape and variability when leveraging artificial intelligence in healthcare to aid medical practitioners. Nonetheless, the development of a tailored dataset for object detection, like body packs, requires careful curation by subject matter specialists to ensure successful training.
METHODS: In this cross-sectional study, abdominal CT scan images were employed to create a machine learning-based model for detecting body packs. A single-step object detection called RetinaNet using a modified neck (Proposed Model) was performed to achieve the best results. Also, an angled Bbox (oriented bounding box) in the training dataset played an important role in improving the results.
RESULTS: A total of 888 abdominal CT scan images were studied. Our proposed Body Packs Detection (BPD) model achieved a mean average precision (mAP) value of 86.6% when the intersection over union (IoU) was 0.5, and a mAP value of 45.6% at different IoU thresholds (from 0.5 to 0.95 in steps of 0.05). It also obtained a Recall value of 58.5%, which was the best result among the standard object detection methods such as the standard RetinaNet.
CONCLUSION: This study employed a deep learning network to identify body packs in abdominal CT scans, highlighting the importance of incorporating object shape and variability when leveraging artificial intelligence in healthcare to aid medical practitioners. Nonetheless, the development of a tailored dataset for object detection, like body packs, requires careful curation by subject matter specialists to ensure successful training.