Automated Detection of Pancreatic Cystic Lesions on CT Using Deep Learning
Pancreatic cystic lesions (PCL) are a frequent and underreported incidental finding on CT scans and can transform into neoplasms with devastating consequences. We developed and evaluated an algorithm based on a two-step nnU-Net architecture for automated detection of PCL on CTs. A total of 543 cysts...
| Main Authors: | , , , , , , , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2021-05-01
|
| Series: | Diagnostics |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2075-4418/11/5/901 |
| _version_ | 1797533567236964352 |
|---|---|
| author | Lorraine Abel Jakob Wasserthal Thomas Weikert Alexander W. Sauter Ivan Nesic Marko Obradovic Shan Yang Sebastian Manneck Carl Glessgen Johanna M. Ospel Bram Stieltjes Daniel T. Boll Björn Friebe |
| author_facet | Lorraine Abel Jakob Wasserthal Thomas Weikert Alexander W. Sauter Ivan Nesic Marko Obradovic Shan Yang Sebastian Manneck Carl Glessgen Johanna M. Ospel Bram Stieltjes Daniel T. Boll Björn Friebe |
| author_sort | Lorraine Abel |
| collection | DOAJ |
| description | Pancreatic cystic lesions (PCL) are a frequent and underreported incidental finding on CT scans and can transform into neoplasms with devastating consequences. We developed and evaluated an algorithm based on a two-step nnU-Net architecture for automated detection of PCL on CTs. A total of 543 cysts on 221 abdominal CTs were manually segmented in 3D by a radiology resident in consensus with a board-certified radiologist specialized in abdominal radiology. This information was used to train a two-step nnU-Net for detection with the performance assessed depending on lesions’ volume and location in comparison to three human readers of varying experience. Mean sensitivity was 78.8 ± 0.1%. The sensitivity was highest for large lesions with 87.8% for cysts ≥220 mm<sup>3</sup> and for lesions in the distal pancreas with up to 96.2%. The number of false-positive detections for cysts ≥220 mm<sup>3</sup> was 0.1 per case. The algorithm’s performance was comparable to human readers. To conclude, automated detection of PCL on CTs is feasible. The proposed model could serve radiologists as a second reading tool. All imaging data and code used in this study are freely available online. |
| first_indexed | 2024-03-10T11:17:22Z |
| format | Article |
| id | doaj.art-adcd95dab082416face4eea33cc27e47 |
| institution | Directory Open Access Journal |
| issn | 2075-4418 |
| language | English |
| last_indexed | 2024-03-10T11:17:22Z |
| publishDate | 2021-05-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Diagnostics |
| spelling | doaj.art-adcd95dab082416face4eea33cc27e472023-11-21T20:20:39ZengMDPI AGDiagnostics2075-44182021-05-0111590110.3390/diagnostics11050901Automated Detection of Pancreatic Cystic Lesions on CT Using Deep LearningLorraine Abel0Jakob Wasserthal1Thomas Weikert2Alexander W. Sauter3Ivan Nesic4Marko Obradovic5Shan Yang6Sebastian Manneck7Carl Glessgen8Johanna M. Ospel9Bram Stieltjes10Daniel T. Boll11Björn Friebe12Clinic of Radiology & Nuclear Medicine, University Hospital Basel, University of Basel, Petersgraben 4, 4031 Basel, SwitzerlandClinic of Radiology & Nuclear Medicine, University Hospital Basel, University of Basel, Petersgraben 4, 4031 Basel, SwitzerlandClinic of Radiology & Nuclear Medicine, University Hospital Basel, University of Basel, Petersgraben 4, 4031 Basel, SwitzerlandClinic of Radiology & Nuclear Medicine, University Hospital Basel, University of Basel, Petersgraben 4, 4031 Basel, SwitzerlandClinic of Radiology & Nuclear Medicine, University Hospital Basel, University of Basel, Petersgraben 4, 4031 Basel, SwitzerlandClinic of Radiology & Nuclear Medicine, University Hospital Basel, University of Basel, Petersgraben 4, 4031 Basel, SwitzerlandClinic of Radiology & Nuclear Medicine, University Hospital Basel, University of Basel, Petersgraben 4, 4031 Basel, SwitzerlandClinic of Radiology & Nuclear Medicine, University Hospital Basel, University of Basel, Petersgraben 4, 4031 Basel, SwitzerlandClinic of Radiology & Nuclear Medicine, University Hospital Basel, University of Basel, Petersgraben 4, 4031 Basel, SwitzerlandClinic of Radiology & Nuclear Medicine, University Hospital Basel, University of Basel, Petersgraben 4, 4031 Basel, SwitzerlandClinic of Radiology & Nuclear Medicine, University Hospital Basel, University of Basel, Petersgraben 4, 4031 Basel, SwitzerlandClinic of Radiology & Nuclear Medicine, University Hospital Basel, University of Basel, Petersgraben 4, 4031 Basel, SwitzerlandClinic of Radiology & Nuclear Medicine, University Hospital Basel, University of Basel, Petersgraben 4, 4031 Basel, SwitzerlandPancreatic cystic lesions (PCL) are a frequent and underreported incidental finding on CT scans and can transform into neoplasms with devastating consequences. We developed and evaluated an algorithm based on a two-step nnU-Net architecture for automated detection of PCL on CTs. A total of 543 cysts on 221 abdominal CTs were manually segmented in 3D by a radiology resident in consensus with a board-certified radiologist specialized in abdominal radiology. This information was used to train a two-step nnU-Net for detection with the performance assessed depending on lesions’ volume and location in comparison to three human readers of varying experience. Mean sensitivity was 78.8 ± 0.1%. The sensitivity was highest for large lesions with 87.8% for cysts ≥220 mm<sup>3</sup> and for lesions in the distal pancreas with up to 96.2%. The number of false-positive detections for cysts ≥220 mm<sup>3</sup> was 0.1 per case. The algorithm’s performance was comparable to human readers. To conclude, automated detection of PCL on CTs is feasible. The proposed model could serve radiologists as a second reading tool. All imaging data and code used in this study are freely available online.https://www.mdpi.com/2075-4418/11/5/901pancreatic cystic lesionintraductal papillary mucinous neoplasiatomographyX-ray computeddetectionartificial intelligence |
| spellingShingle | Lorraine Abel Jakob Wasserthal Thomas Weikert Alexander W. Sauter Ivan Nesic Marko Obradovic Shan Yang Sebastian Manneck Carl Glessgen Johanna M. Ospel Bram Stieltjes Daniel T. Boll Björn Friebe Automated Detection of Pancreatic Cystic Lesions on CT Using Deep Learning Diagnostics pancreatic cystic lesion intraductal papillary mucinous neoplasia tomography X-ray computed detection artificial intelligence |
| title | Automated Detection of Pancreatic Cystic Lesions on CT Using Deep Learning |
| title_full | Automated Detection of Pancreatic Cystic Lesions on CT Using Deep Learning |
| title_fullStr | Automated Detection of Pancreatic Cystic Lesions on CT Using Deep Learning |
| title_full_unstemmed | Automated Detection of Pancreatic Cystic Lesions on CT Using Deep Learning |
| title_short | Automated Detection of Pancreatic Cystic Lesions on CT Using Deep Learning |
| title_sort | automated detection of pancreatic cystic lesions on ct using deep learning |
| topic | pancreatic cystic lesion intraductal papillary mucinous neoplasia tomography X-ray computed detection artificial intelligence |
| url | https://www.mdpi.com/2075-4418/11/5/901 |
| work_keys_str_mv | AT lorraineabel automateddetectionofpancreaticcysticlesionsonctusingdeeplearning AT jakobwasserthal automateddetectionofpancreaticcysticlesionsonctusingdeeplearning AT thomasweikert automateddetectionofpancreaticcysticlesionsonctusingdeeplearning AT alexanderwsauter automateddetectionofpancreaticcysticlesionsonctusingdeeplearning AT ivannesic automateddetectionofpancreaticcysticlesionsonctusingdeeplearning AT markoobradovic automateddetectionofpancreaticcysticlesionsonctusingdeeplearning AT shanyang automateddetectionofpancreaticcysticlesionsonctusingdeeplearning AT sebastianmanneck automateddetectionofpancreaticcysticlesionsonctusingdeeplearning AT carlglessgen automateddetectionofpancreaticcysticlesionsonctusingdeeplearning AT johannamospel automateddetectionofpancreaticcysticlesionsonctusingdeeplearning AT bramstieltjes automateddetectionofpancreaticcysticlesionsonctusingdeeplearning AT danieltboll automateddetectionofpancreaticcysticlesionsonctusingdeeplearning AT bjornfriebe automateddetectionofpancreaticcysticlesionsonctusingdeeplearning |