Artificial Intelligence Accurately Detects Traumatic Thoracolumbar Fractures on Sagittal Radiographs
<i>Background and Objectives</i>: Commonly being the first step in trauma routine imaging, up to 67% fractures are missed on plain radiographs of the thoracolumbar (TL) spine. The aim of this study was to develop a deep learning model that detects traumatic fractures on sagittal radiogra...
| Main Authors: | , , , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2022-07-01
|
| Series: | Medicina |
| Subjects: | |
| Online Access: | https://www.mdpi.com/1648-9144/58/8/998 |
| _version_ | 1827599108205969408 |
|---|---|
| author | Guillermo Sánchez Rosenberg Andrea Cina Giuseppe Rosario Schiró Pietro Domenico Giorgi Boyko Gueorguiev Mauro Alini Peter Varga Fabio Galbusera Enrico Gallazzi |
| author_facet | Guillermo Sánchez Rosenberg Andrea Cina Giuseppe Rosario Schiró Pietro Domenico Giorgi Boyko Gueorguiev Mauro Alini Peter Varga Fabio Galbusera Enrico Gallazzi |
| author_sort | Guillermo Sánchez Rosenberg |
| collection | DOAJ |
| description | <i>Background and Objectives</i>: Commonly being the first step in trauma routine imaging, up to 67% fractures are missed on plain radiographs of the thoracolumbar (TL) spine. The aim of this study was to develop a deep learning model that detects traumatic fractures on sagittal radiographs of the TL spine. Identifying vertebral fractures in simple radiographic projections would have a significant clinical and financial impact, especially for low- and middle-income countries where computed tomography (CT) and magnetic resonance imaging (MRI) are not readily available and could help select patients that need second level imaging, thus improving the cost-effectiveness. <i>Materials and Methods</i>: Imaging studies (radiographs, CT, and/or MRI) of 151 patients were used. An expert group of three spinal surgeons reviewed all available images to confirm presence and type of fractures. In total, 630 single vertebra images were extracted from the sagittal radiographs of the 151 patients—302 exhibiting a vertebral body fracture, and 328 exhibiting no fracture. Following augmentation, these single vertebra images were used to train, validate, and comparatively test two deep learning convolutional neural network models, namely ResNet18 and VGG16. A heatmap analysis was then conducted to better understand the predictions of each model. <i>Results</i>: ResNet18 demonstrated a better performance, achieving higher sensitivity (91%), specificity (89%), and accuracy (88%) compared to VGG16 (90%, 83%, 86%). In 81% of the cases, the “warm zone” in the heatmaps correlated with the findings, suggestive of fracture within the vertebral body seen in the imaging studies. Vertebras T12 to L2 were the most frequently involved, accounting for 48% of the fractures. A4, A3, and A1 were the most frequent fracture types according to the AO Spine Classification. <i>Conclusions</i>: ResNet18 could accurately identify the traumatic vertebral fractures on the TL sagittal radiographs. In most cases, the model based its prediction on the same areas that human expert classifiers used to determine the presence of a fracture. |
| first_indexed | 2024-03-09T04:07:55Z |
| format | Article |
| id | doaj.art-99bf520e60884cf0becea9ffb588d8a0 |
| institution | Directory Open Access Journal |
| issn | 1010-660X 1648-9144 |
| language | English |
| last_indexed | 2024-03-09T04:07:55Z |
| publishDate | 2022-07-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Medicina |
| spelling | doaj.art-99bf520e60884cf0becea9ffb588d8a02023-12-03T14:04:22ZengMDPI AGMedicina1010-660X1648-91442022-07-0158899810.3390/medicina58080998Artificial Intelligence Accurately Detects Traumatic Thoracolumbar Fractures on Sagittal RadiographsGuillermo Sánchez Rosenberg0Andrea Cina1Giuseppe Rosario Schiró2Pietro Domenico Giorgi3Boyko Gueorguiev4Mauro Alini5Peter Varga6Fabio Galbusera7Enrico Gallazzi8AO Research Institute Davos, 7270 Davos, SwitzerlandIRCCS Istituto Ortopedico Galeazzi, 20161 Milano, ItalyASST GOM Niguarda, 20161 Milano, ItalyASST GOM Niguarda, 20161 Milano, ItalyAO Research Institute Davos, 7270 Davos, SwitzerlandAO Research Institute Davos, 7270 Davos, SwitzerlandAO Research Institute Davos, 7270 Davos, SwitzerlandSpine Center, Schulthess Clinic, 8008 Zurich, SwitzerlandUOC Patologia Vertebrale e Scoliosi, ASST Gaetano Pini-CTO, 20161 Milano, Italy<i>Background and Objectives</i>: Commonly being the first step in trauma routine imaging, up to 67% fractures are missed on plain radiographs of the thoracolumbar (TL) spine. The aim of this study was to develop a deep learning model that detects traumatic fractures on sagittal radiographs of the TL spine. Identifying vertebral fractures in simple radiographic projections would have a significant clinical and financial impact, especially for low- and middle-income countries where computed tomography (CT) and magnetic resonance imaging (MRI) are not readily available and could help select patients that need second level imaging, thus improving the cost-effectiveness. <i>Materials and Methods</i>: Imaging studies (radiographs, CT, and/or MRI) of 151 patients were used. An expert group of three spinal surgeons reviewed all available images to confirm presence and type of fractures. In total, 630 single vertebra images were extracted from the sagittal radiographs of the 151 patients—302 exhibiting a vertebral body fracture, and 328 exhibiting no fracture. Following augmentation, these single vertebra images were used to train, validate, and comparatively test two deep learning convolutional neural network models, namely ResNet18 and VGG16. A heatmap analysis was then conducted to better understand the predictions of each model. <i>Results</i>: ResNet18 demonstrated a better performance, achieving higher sensitivity (91%), specificity (89%), and accuracy (88%) compared to VGG16 (90%, 83%, 86%). In 81% of the cases, the “warm zone” in the heatmaps correlated with the findings, suggestive of fracture within the vertebral body seen in the imaging studies. Vertebras T12 to L2 were the most frequently involved, accounting for 48% of the fractures. A4, A3, and A1 were the most frequent fracture types according to the AO Spine Classification. <i>Conclusions</i>: ResNet18 could accurately identify the traumatic vertebral fractures on the TL sagittal radiographs. In most cases, the model based its prediction on the same areas that human expert classifiers used to determine the presence of a fracture.https://www.mdpi.com/1648-9144/58/8/998vertebral fracturefracture detectionheatmapmachine learningartificial intelligence |
| spellingShingle | Guillermo Sánchez Rosenberg Andrea Cina Giuseppe Rosario Schiró Pietro Domenico Giorgi Boyko Gueorguiev Mauro Alini Peter Varga Fabio Galbusera Enrico Gallazzi Artificial Intelligence Accurately Detects Traumatic Thoracolumbar Fractures on Sagittal Radiographs Medicina vertebral fracture fracture detection heatmap machine learning artificial intelligence |
| title | Artificial Intelligence Accurately Detects Traumatic Thoracolumbar Fractures on Sagittal Radiographs |
| title_full | Artificial Intelligence Accurately Detects Traumatic Thoracolumbar Fractures on Sagittal Radiographs |
| title_fullStr | Artificial Intelligence Accurately Detects Traumatic Thoracolumbar Fractures on Sagittal Radiographs |
| title_full_unstemmed | Artificial Intelligence Accurately Detects Traumatic Thoracolumbar Fractures on Sagittal Radiographs |
| title_short | Artificial Intelligence Accurately Detects Traumatic Thoracolumbar Fractures on Sagittal Radiographs |
| title_sort | artificial intelligence accurately detects traumatic thoracolumbar fractures on sagittal radiographs |
| topic | vertebral fracture fracture detection heatmap machine learning artificial intelligence |
| url | https://www.mdpi.com/1648-9144/58/8/998 |
| work_keys_str_mv | AT guillermosanchezrosenberg artificialintelligenceaccuratelydetectstraumaticthoracolumbarfracturesonsagittalradiographs AT andreacina artificialintelligenceaccuratelydetectstraumaticthoracolumbarfracturesonsagittalradiographs AT giusepperosarioschiro artificialintelligenceaccuratelydetectstraumaticthoracolumbarfracturesonsagittalradiographs AT pietrodomenicogiorgi artificialintelligenceaccuratelydetectstraumaticthoracolumbarfracturesonsagittalradiographs AT boykogueorguiev artificialintelligenceaccuratelydetectstraumaticthoracolumbarfracturesonsagittalradiographs AT mauroalini artificialintelligenceaccuratelydetectstraumaticthoracolumbarfracturesonsagittalradiographs AT petervarga artificialintelligenceaccuratelydetectstraumaticthoracolumbarfracturesonsagittalradiographs AT fabiogalbusera artificialintelligenceaccuratelydetectstraumaticthoracolumbarfracturesonsagittalradiographs AT enricogallazzi artificialintelligenceaccuratelydetectstraumaticthoracolumbarfracturesonsagittalradiographs |