Deep Learning-Based Computer-Aided Diagnosis in Coronary Artery Calcium-Scoring CT for Pulmonary Nodule Detection: A Preliminary Study

Abstract

Purpose: To evaluate the feasibility and utility of deep learning-based computer-aided diagnosis (DL-CAD) for the detection of pulmonary nodules on coronary artery calcium (CAC)-scoring computed tomography (CT).

Materials and methods: This retrospective study included 273 patients (aged 63.9±13.2 years; 129 men) who underwent CAC-scoring CT. A DL-CAD system based on thin-section images was used for pulmonary nodule detection, and two independent junior readers reviewed the standard CAC-scoring CT scans with and without referencing DL-CAD results. A reference standard was established through the consensus of two experienced radiologists. Sensitivity, positive predictive value, and F1-score were assessed on a per-nodule and per-patient basis. The patients' medical records were monitored until November 2023.

Results: A total of 269 nodules were identified in 129 patients. With DL-CAD assistance, the readers' sensitivity significantly improved (65% vs. 80% for reader 1; 82% vs. 86% for reader 2; all p<0.001), without a notable increase in the number of false-positives per case (0.11 vs. 0.13, p=0.078 for reader 1; 0.11 vs. 0.11, p>0.999 for reader 2). Per-patient analysis also enhanced sensitivity with DL-CAD assistance (73% vs. 84%, p<0.001 for reader 1; 89% vs. 91%, p=0.250 for reader 2). During follow-up, lung cancer was diagnosed in four patients (1.5%). Among them, two had lesions detected on CAC-scoring CT, both of which were successfully identified by DL-CAD.

Conclusion: DL-CAD based on thin-section images can assist less experienced readers in detecting pulmonary nodules on CAC-scoring CT scans, improving detection sensitivity without significantly increasing false-positives.

Keywords: Computer-aided diagnosis; X-ray computed; deep learning; diagnostic performance; lung; tomography.

Fig. 1. Flow diagram of study participants. CABG, coronary artery bypass graft; CAC, coronary artery calcium; CAD, computer-aided diagnosis; CT, computed tomography; TAVR, transcatheter aortic valve replacement.

Fig. 2. Comprehensive evaluation of CAC-scoring CT utilizing the full field-of-view. In addition to the conventional interpretation using 3-mm soft kernel images, DL-CAD based on thin-section sharp kernel images was used to assist readers in detecting lung nodules. CAC, coronary artery calcium; CT, computed tomography; DL-CAD, deep learning-based computer-aided diagnosis.

Fig. 3. Detection sensitivity of DL-CAD and readers without and with the assistance of DL-CAD for nodules (269 nodules in 129 patients). Assistance of DL-CAD improved detection sensitivity in both per-nodule and per-patient analysis in readers. DL-CAD, deep learning-based computer-aided diagnosis.

Fig. 4. Examples of correct DL-CAD detection of pulmonary nodules. (A) Lobulated pure ground-glass nodule in the right middle lob, resected and confirmed as minimally invasive adenocarcinoma. Both readers also correctly identified this lesion. (B) Part-solid nodule with spiculation in the right lower lobe, highly suggestive of lung cancer. While reader 1 detected the lesion, reader 2 showed a false-negative, and this result remained unchanged after referencing DL-CAD results. (C and D) Successful DL-CAD detection of small nodules near bronchovascular bundles, missed by both readers (C) and by reader 1 (D). After reviewing the DL-CAD results, both lesions were converted to true-positives by reader 1. (E and F) DL-CAD identifies nodules abutting costal pleura, overlooked by reader 1 (E) and reader 2 (F). The reader assessment did not change after referring DL-CAD results. DL-CAD, deep learning-based computer-aided diagnosis.