Abstract
The diagnostic performance of CT for pancreatic cancer is interpreter-dependent, and approximately 40% of tumours smaller than 2 cm evade detection. Convolutional neural networks (CNNs) have shown promise in image analysis, but the networks' potential for pancreatic cancer detection and diagnosis is unclear. We aimed to investigate whether CNN could distinguish individuals with and without pancreatic cancer on CT, compared with radiologist interpretation. In this retrospective, diagnostic study, contrast-enhanced CT images of 370 patients with pancreatic cancer and 320 controls from a Taiwanese centre were manually labelled and randomly divided for training and validation (295 patients with pancreatic cancer and 256 controls) and testing (75 patients with pancreatic cancer and 64 controls; local test set 1). Images were preprocessed into patches, and a CNN was trained to classify patches as cancerous or non-cancerous. Individuals were classified as with or without pancreatic cancer on the basis of the proportion of patches diagnosed as cancerous by the CNN, using a cutoff determined using the training and validation set. The CNN was further tested with another local test set (101 patients with pancreatic cancers and 88 controls; local test set 2) and a US dataset (281 pancreatic cancers and 82 controls). Radiologist reports of pancreatic cancer images in the local test sets were retrieved for comparison. Between Jan 1, 2006, and Dec 31, 2018, we obtained CT images. In local test set 1, CNN-based analysis had a sensitivity of 0·973, specificity of 1·000, and accuracy of 0·986 (area under the curve [AUC] 0·997 (95% CI 0·992-1·000). In local test set 2, CNN-based analysis had a sensitivity of 0·990, specificity of 0·989, and accuracy of 0·989 (AUC 0·999 [0·998-1·000]). In the US test set, CNN-based analysis had a sensitivity of 0·790, specificity of 0·976, and accuracy of 0·832 (AUC 0·920 [0·891-0·948)]. CNN-based analysis achieved higher sensitivity than radiologists did (0·983 vs 0·929, difference 0·054 [95% CI 0·011-0·098]; p=0·014) in the two local test sets combined. CNN missed three (1·7%) of 176 pancreatic cancers (1·1-1·2 cm). Radiologists missed 12 (7%) of 168 pancreatic cancers (1·0-3·3 cm), of which 11 (92%) were correctly classified using CNN. The sensitivity of CNN for tumours smaller than 2 cm was 92·1% in the local test sets and 63·1% in the US test set. CNN could accurately distinguish pancreatic cancer on CT, with acceptable generalisability to images of patients from various races and ethnicities. CNN could supplement radiologist interpretation. Taiwan Ministry of Science and Technology.
Highlights
Pancreatic cancer is projected to become the second leading cause of cancer deaths in the USA by 2030,1 with dismal survival once the tumour size exceeds 2 cm.[2,3] CT is the major imaging modality used for detection and assessment of pancreatic cancer,[4] but the method’s diag nostic performance depends on radiologists’ experience
CT image datasets and radiologist reports Patients with histologically confirmed or cytologically confirmed pancreatic adenocarcinoma were identified from the National Taiwan University Hospital (NTUH) Cancer Registry, and the CT images of those patients obtained before the date of pancreatic cancer diagnosis were extracted from the imaging archive of NTUH for review
Contrast-enhanced portal venous CT images of 370 pancreatic cancer patients diagnosed between Jan 1, 2013, and Dec 31, 2018, and 320 patients with normal pancreas during the same period were randomly selected for training and testing of the deep learning model
Summary
Pancreatic cancer is projected to become the second leading cause of cancer deaths in the USA by 2030,1 with dismal survival once the tumour size exceeds 2 cm.[2,3] CT is the major imaging modality used for detection and assessment of pancreatic cancer,[4] but the method’s diag nostic performance depends on radiologists’ experience. Approximately 40% of tumours that are smaller than 2 cm evade detection by CT,[5] underscoring an urgent need for novel methods to supplement radiologist interpretation in improving the sensitivity for the detection of pancreatic cancer. CNN has been reported to achieve a high accuracy in the imaging diagnosis of various conditions including skin cancer,[7] diabetic retinopathy,[8] and liver masses.[9] the potential usefulness of CNN for the detection and diagnosis of pancreatic cancer has not been widely investigated. Most pancreatic cancers present with irregular contours and ill-defined margins on CT and are often obscure at an early stage, posing substantial challenges even for the most experienced radiologists.[10,11] www.thelancet.com/digital-health Vol 2 June 2020
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
