Using AI to Improve Radiologist Performance in Detection of Abnormalities on Chest Radiographs.

Background Chest radiography remains the most common radiologic examination, and interpretation of its results can be difficult. Purpose To explore the potential benefit of artificial intelligence (AI) assistance in the detection of thoracic abnormalities on chest radiographs by evaluating the performance of radiologists with different levels of expertise, with and without AI assistance. Materials and Methods Patients who underwent both chest radiography and thoracic CT within 72 hours between January 2010 and December 2020 in a French public hospital were screened retrospectively. Radiographs were randomly included until reaching 500 radiographs, with about 50% of radiographs having abnormal findings. A senior thoracic radiologist annotated the radiographs for five abnormalities (pneumothorax, pleural effusion, consolidation, mediastinal and hilar mass, lung nodule) based on the corresponding CT results (ground truth). A total of 12 readers (four thoracic radiologists, four general radiologists, four radiology residents) read half the radiographs without AI and half the radiographs with AI (ChestView; Gleamer). Changes in sensitivity and specificity were measured using paired t tests. Results The study included 500 patients (mean age, 54 years ± 19 [SD]; 261 female, 239 male), with 522 abnormalities visible on 241 radiographs. On average, for all readers, AI use resulted in an absolute increase in sensitivity of 26% (95% CI: 20, 32), 14% (95% CI: 11, 17), 12% (95% CI: 10, 14), 8.5% (95% CI: 6, 11), and 5.9% (95% CI: 4, 8) for pneumothorax, consolidation, nodule, pleural effusion, and mediastinal and hilar mass, respectively (P < .001). Specificity increased with AI assistance (3.9% [95% CI: 3.2, 4.6], 3.7% [95% CI: 3, 4.4], 2.9% [95% CI: 2.3, 3.5], and 2.1% [95% CI: 1.6, 2.6] for pleural effusion, mediastinal and hilar mass, consolidation, and nodule, respectively), except in the diagnosis of pneumothorax (-0.2%; 95% CI: -0.36, -0.04; P = .01). The mean reading time was 81 seconds without AI versus 56 seconds with AI (31% decrease, P < .001). Conclusion AI-assisted chest radiography interpretation resulted in absolute increases in sensitivity for all radiologists of various levels of expertise and reduced the reading times; specificity increased with AI, except in the diagnosis of pneumothorax. © RSNA, 2023 Supplemental material is available for this article.

AI-driven quantification, staging and outcome prediction of COVID-19 pneumonia.

Coronavirus disease 2019 (COVID-19) emerged in 2019 and disseminated around the world rapidly. Computed tomography (CT) imaging has been proven to be an important tool for screening, disease quantification and staging. The latter is of extreme importance for organizational anticipation (availability of intensive care unit beds, patient management planning) as well as to accelerate drug development through rapid, reproducible and quantified assessment of treatment response. Even if currently there are no specific guidelines for the staging of the patients, CT together with some clinical and biological biomarkers are used. In this study, we collected a multi-center cohort and we investigated the use of medical imaging and artificial intelligence for disease quantification, staging and outcome prediction. Our approach relies on automatic deep learning-based disease quantification using an ensemble of architectures, and a data-driven consensus for the staging and outcome prediction of the patients fusing imaging biomarkers with clinical and biological attributes. Highly promising results on multiple external/independent evaluation cohorts as well as comparisons with expert human readers demonstrate the potentials of our approach.

Recurrent acute pancreatitis caused by combined gastric and pancreatic duplications: value of preoperative CT and MRCP and conservative surgery.

We present a series of patients with recurrent acute pancreatitis caused by a duplicated pancreatic head connected to a gastric duplication and successfully treated by conservative surgery. This retrospective study included consecutive adult patients referred to our institution for recurrent acute pancreatitis. All patients underwent a preoperative non-invasive imaging examination including contrast-enhanced computed tomography and magnetic resonance cholangiopancreatography (MRCP). The final diagnosis of this developmental anomaly was based on surgical and pathological examinations. The four patients in this study had the same typical imaging pattern including a duplicated duct. There was no recurrent acute pancreatitis after surgical treatment, which involved atypical resection of the duplicated pancreatic head and segmental gastric resection, without a Whipple procedure. The discovery of an accessory pancreatic head with a duct terminating in a cyst identified on MRCP in a patient with recurrent acute pancreatitis could suggest this rare and surgically treatable cause of acute pancreatitis.

Non-measurable infiltrative HCC: is post-contrast attenuation on CT a sign of tumor response?

OBJECTIVES: To evaluate the value of CT attenuation to assess the response to sorafenib in infiltrative/endovascular non-measurable advanced hepatocellular carcinoma (HCC). METHODS: From 2007 to 2014, patients with infiltrative HCC ± tumor-in-vein (TIV) were retrospectively included. Attenuation of tumors and TIV were measured at baseline and follow-up on arterial and portal venous phase CT by two independent radiologists. Attenuation changes (overall and as per Choi criteria) and Child-Pugh score were correlated to overall survival. RESULTS: Forty patients were included (38 men, 95%). Attenuation of both the tumors and TIV was significantly lower in follow-up CT than on baseline CT (p = 0.002 (arterial), and p = 0.001 (portal) for tumor, and p = 0.004 (arterial) and p < 0.001 (porta) for TIV). Median attenuation of TIV was significantly lower than that of the tumor in follow-up images (p = 0.010). Median OS for the entire cohort was 4 ± 1 months (95% CI: 2.1-5.9), with estimated OS rates at 6, 12, and 24 months of 43%, 29 and 12%, respectively. Baseline and follow-up CT attenuation in tumors and TVI were not correlated with survival. Survival was not significantly increased in patients with Choi criteria >15% CT HU decrease in the tumor and/or TIV during follow-up. Only Child-Pugh A (HR 4.9 (95%CI 2.3-10.7), p < 0.001) was identified as an independent factor of improved survival on multivariate analysis. CONCLUSION: Despite significant changes under sorafenib, tumor attenuation of infiltrative/endovascular non-measurable HCC may be of limited value to assess survival in this subgroup of patients with very poor prognosis. KEY POINTS: • Attenuation of both tumors and tumor-in-vein decreases after sorafenib. • Attenuation decrease is more marked in the tumor-in-vein than in the tumor. • Attenuation decrease is not associated with longer overall survival.

DWI-ASPECTS (Diffusion-Weighted Imaging-Alberta Stroke Program Early Computed Tomography Scores) and DWI-FLAIR (Diffusion-Weighted Imaging-Fluid Attenuated Inversion Recovery) Mismatch in Thrombectomy Candidates: An Intrarater and Interrater Agreement Study.

BACKGROUND AND PURPOSE: We aimed to study the intrarater and interrater agreement of clinicians attributing DWI-ASPECTS (Diffusion-Weighted Imaging-Alberta Stroke Program Early Computed Tomography Scores) and DWI-FLAIR (Diffusion-Weighted Imaging-Fluid Attenuated Inversion Recovery) mismatch in patients with acute ischemic stroke referred for mechanical thrombectomy. METHODS: Eighteen raters independently scored anonymized magnetic resonance imaging scans of 30 participants from a multicentre thrombectomy trial, in 2 different reading sessions. Agreement was measured using Fleiss κ and Cohen κ statistics. RESULTS: Interrater agreement for DWI-ASPECTS was slight (κ=0.17 [0.14-0.21]). Four raters (22.2%) had a substantial (or higher) intrarater agreement. Dichotomization of the DWI-ASPECTS (0-5 versus 6-10 or 0-6 versus 7-10) increased the interrater agreement to a substantial level (κ=0.62 [0.48-0.75] and 0.68 [0.55-0.79], respectively) and more raters reached a substantial (or higher) intrarater agreement (17/18 raters [94.4%]). Interrater agreement for DWI-FLAIR mismatch was moderate (κ=0.43 [0.33-0.57]); 11 raters (61.1%) reached a substantial (or higher) intrarater agreement. CONCLUSIONS: Agreement between clinicians assessing DWI-ASPECTS and DWI-FLAIR mismatch may not be sufficient to make repeatable clinical decisions in mechanical thrombectomy. The dichotomization of the DWI-ASPECTS (0-5 versus 0-6 or 0-6 versus 7-10) improved interrater and intrarater agreement, however, its relevance for patients selection for mechanical thrombectomy needs to be validated in a randomized trial.