How artificial intelligence improves radiological interpretation in suspected pulmonary embolism.

OBJECTIVES: To evaluate and compare the diagnostic performances of a commercialized artificial intelligence (AI) algorithm for diagnosing pulmonary embolism (PE) on CT pulmonary angiogram (CTPA) with those of emergency radiologists in routine clinical practice. METHODS: This was an IRB-approved retrospective multicentric study including patients with suspected PE from September to December 2019 (i.e., during a preliminary evaluation period of an approved AI algorithm). CTPA quality and conclusions by emergency radiologists were retrieved from radiological reports. The gold standard was a retrospective review of CTPA, radiological and clinical reports, AI outputs, and patient outcomes. Diagnostic performance metrics for AI and radiologists were assessed in the entire cohort and depending on CTPA quality. RESULTS: Overall, 1202 patients were included (median age: 66.2 years). PE prevalence was 15.8% (190/1202). The AI algorithm detected 219 suspicious PEs, of which 176 were true PEs, including 19 true PEs missed by radiologists. In the cohort, the highest sensitivity and negative predictive values (NPVs) were obtained with AI (92.6% versus 90% and 98.6% versus 98.1%, respectively), while the highest specificity and positive predictive value (PPV) were found with radiologists (99.1% versus 95.8% and 95% versus 80.4%, respectively). Accuracy, specificity, and PPV were significantly higher for radiologists except in subcohorts with poor-to-average injection quality. Radiologists positively evaluated the AI algorithm to improve their diagnostic comfort (55/79 [69.6%]). CONCLUSION: Instead of replacing radiologists, AI for PE detection appears to be a safety net in emergency radiology practice due to high sensitivity and NPV, thereby increasing the self-confidence of radiologists. KEY POINTS: • Both the AI algorithm and emergency radiologists showed excellent performance in diagnosing PE on CTPA (sensitivity and specificity ≥ 90%; accuracy ≥ 95%). • The AI algorithm for PE detection can help increase the sensitivity and NPV of emergency radiologists in clinical practice, especially in cases of poor-to-moderate injection quality. • Emergency radiologists recommended the use of AI for PE detection in satisfaction surveys to increase their confidence and comfort in their final diagnosis.

Practical clinical and radiological models to diagnose COVID-19 based on a multicentric teleradiological emergency chest CT cohort.

Our aim was to develop practical models built with simple clinical and radiological features to help diagnosing Coronavirus disease 2019 [COVID-19] in a real-life emergency cohort. To do so, 513 consecutive adult patients suspected of having COVID-19 from 15 emergency departments from 2020-03-13 to 2020-04-14 were included as long as chest CT-scans and real-time polymerase chain reaction (RT-PCR) results were available (244 [47.6%] with a positive RT-PCR). Immediately after their acquisition, the chest CTs were prospectively interpreted by on-call teleradiologists (OCTRs) and systematically reviewed within one week by another senior teleradiologist. Each OCTR reading was concluded using a 5-point scale: normal, non-infectious, infectious non-COVID-19, indeterminate and highly suspicious of COVID-19. The senior reading reported the lesions' semiology, distribution, extent and differential diagnoses. After pre-filtering clinical and radiological features through univariate Chi-2, Fisher or Student t-tests (as appropriate), multivariate stepwise logistic regression (Step-LR) and classification tree (CART) models to predict a positive RT-PCR were trained on 412 patients, validated on an independent cohort of 101 patients and compared with the OCTR performances (295 and 71 with available clinical data, respectively) through area under the receiver operating characteristics curves (AUC). Regarding models elaborated on radiological variables alone, best performances were reached with the CART model (i.e., AUC = 0.92 [versus 0.88 for OCTR], sensitivity = 0.77, specificity = 0.94) while step-LR provided the highest AUC with clinical-radiological variables (AUC = 0.93 [versus 0.86 for OCTR], sensitivity = 0.82, specificity = 0.91). Hence, these two simple models, depending on the availability of clinical data, provided high performances to diagnose positive RT-PCR and could be used by any radiologist to support, modulate and communicate their conclusion in case of COVID-19 suspicion. Practically, using clinical and radiological variables (GGO, fever, presence of fibrotic bands, presence of diffuse lesions, predominant peripheral distribution) can accurately predict RT-PCR status.

Location of residual cancer after transrectal high-intensity focused ultrasound ablation for clinically localized prostate cancer.

UNLABELLED: What's known on the subject ? and What does the study add? Transrectal High-Intensity Focused Ultrasound (HIFU) ablation has been used as a minimally invasive treatment for localized prostate cancer for 15 years. Five-year disease-free survival rates of 66-78% have been reported, challenging the results of external-beam radiation therapy. Usually, a 6-mm safety margin is used in the apex to preserve the urinary sphincter and potency. The influence of this 6-mm margin on the results of the treatment has never been assessed. This retrospective study of a cohort of 99 patients who underwent systematic biopsy 3-6 months after HIFU ablation for prostate cancer (with a 6-mm safety margin in the apex) shows that post-HIFU residual cancer is found more frequently in the apex. Therefore, new strategies improving the prostate destruction at the apex while preserving the urinary continence need to be found. OBJECTIVE: • To evaluate whether the location (apex/midgland/base) of prostate cancer influences the risk of incomplete transrectal high-intensity focused ultrasonography (HIFU) ablation. PATIENTS AND METHODS: • We retrospectively studied 99 patients who underwent prostate cancer HIFU ablation (Ablatherm; EDAP, Vaulx-en-Velin, France) with a 6-mm safety margin at the apex, and had systematic biopsies 3-6 months after treatment. • Locations of positive pre- and post-HIFU sextants were compared. • The present study included two analyses. First, sextants negative before and positive after treatment were recoded as positive/positive, hypothesizing that cancer had been missed at pretreatment biopsy. Second, patients with such sextants were excluded. RESULTS: • Pre-HIFU biopsies found cancer in all patients and in 215/594 sextants (36.2%); 55 (25.6%) positive sextants were in the apex, 86 (40%) in the midgland and 74 (34.4%) in the base. • After treatment, residual cancer was found in 36 patients (36.4%) and 50 sextants (8.4%); 30 (60%) positive sextants were in the apex, 12 (24%) in the midgland and eight (16%) in the base. • Both statistical analyses found that the locations of the positive sextants before and after HIFU ablation were significantly different (P < 0.001), with a higher proportion of positive apical sextants after treatment. • At the first analysis, the mean (95% confidence interval) probability for a sextant to remain positive after HIFU ablation was 8.8% (3.5-20.3%) in the base, 12.7% (5.8-25.9%) in the midgland and 41.7% (27.2-57.89%) in the apex. • At the second analysis, these same probabilities were 5.9% (1.9-17%), 9.9% (3.9-23.2%) and 27.3% (13.7-47%), respectively. CONCLUSION: • When a 6-mm apical safety margin is used, residual cancer after HIFU ablation is found significantly more frequently in the apex.

Evaluation of T2-weighted and dynamic contrast-enhanced MRI in localizing prostate cancer before repeat biopsy.

We assessed the accuracy of T2-weighted (T2w) and dynamic contrast-enhanced (DCE) 1.5-T magnetic resonance imaging (MRI) in localizing prostate cancer before transrectal ultrasound-guided repeat biopsy. Ninety-three patients with abnormal PSA level and negative prostate biopsy underwent T2w and DCE prostate MRI using pelvic coil before repeat biopsy. T2w and DCE images were interpreted using visual criteria only. MR results were correlated with repeat biopsy findings in ten prostate sectors. Repeat biopsy found prostate cancer in 23 patients (24.7%) and 44 sectors (6.6%). At per patient analysis, the sensitivity, specificity, positive and negative predictive values were 47.8%, 44.3%, 20.4% and 79.5% for T2w imaging and 82.6%, 20%, 24.4% and 93.3% for DCE imaging. When all suspicious areas (on T2w or DCE imaging) were taken into account, a sensitivity of 82.6% and a negative predictive value of 100% could be achieved. At per sector analysis, DCE imaging was significantly less specific (83.5% vs. 89.7%, p < 0.002) than T2w imaging; it was more sensitive (52.4% vs. 32.1%), but the difference was hardly significant (p = 0.09). T2w and DCE MRI using pelvic coil and visual diagnostic criteria can guide prostate repeat biopsy, with a good sensitivity and NPV.