Development and performance evaluation of a deep learning lung nodule detection system.

BACKGROUND: Lung cancer is the leading cause of cancer-related deaths throughout the world. Chest computed tomography (CT) is now widely used in the screening and diagnosis of lung cancer due to its effectiveness. Radiologists must identify each small nodule shadow from 3D volume images, which is very burdensome and often results in missed nodules. To address these challenges, we developed a computer-aided detection (CAD) system that automatically detects lung nodules in CT images. METHODS: A total of 1997 chest CT scans were collected for algorithm development. The algorithm was designed using deep learning technology. In addition to evaluating detection performance on various public datasets, its robustness to changes in radiation dose was assessed by a phantom study. To investigate the clinical usefulness of the CAD system, a reader study was conducted with 10 doctors, including inexperienced and expert readers. This study investigated whether the use of the CAD as a second reader could prevent nodular lesions in lungs that require follow-up examinations from being overlooked. Analysis was performed using the Jackknife Free-Response Receiver-Operating Characteristic (JAFROC). RESULTS: The CAD system achieved sensitivity of 0.98/0.96 at 3.1/7.25 false positives per case on two public datasets. Sensitivity did not change within the range of practical doses for a study using a phantom. A second reader study showed that the use of this system significantly improved the detection ability of nodules that could be picked up clinically (p = 0.026). CONCLUSIONS: We developed a deep learning-based CAD system that is robust to imaging conditions. Using this system as a second reader increased detection performance.

Clinical validity of non-contrast-enhanced VI-RADS: prospective study using 3-T MRI with high-gradient magnetic field.

OBJECTIVES: To develop a modified Vesical Imaging Reporting and Data System (VI-RADS) without dynamic contrast-enhanced imaging (DCEI), termed "non-contrast-enhanced VI-RADS (NCE-VI-RADS)", and to assess the additive impact of denoising deep learning reconstruction (dDLR) on NCE-VI-RADS. METHODS: From January 2019 through December 2020, 163 participants who underwent high-gradient 3-T MRI of the bladder were prospectively enrolled. In total, 108 participants with pathologically confirmed bladder cancer by transurethral resection were analyzed. Tumors were evaluated based on VI-RADS (scores 1-5) by two readers independently: an experienced radiologist (reader 1) and a senior radiology resident (reader 2). Conventional VI-RADS assessment included all three imaging types (T2-weighted imaging [T2WI], diffusion-weighted imaging [DWI], and dynamic contrast-enhanced imaging [DCEI]). Also evaluated were NCE-VI-RADS comprising only non-contrast-enhanced imaging types (T2WI and DWI), and "NCE-VI-RADS with dDLR" comprising T2WI processed with dDLR and DWI. All systems were assessed using receiver-operating characteristic curve analysis and simple and/or weighted κ statistics. RESULTS: Muscle invasion was identified in 23/108 participants (21%). Area under the curve (AUC) values for diagnosing muscle invasion were as follows: conventional VI-RADS, 0.94 and 0.91; NCE-VI-RADS, 0.93 and 0.91; and "NCE-VI-RADS with dDLR", 0.96 and 0.93, for readers 1 and 2, respectively. Simple κ statistics indicated substantial agreement for NCE-VI-RADS and almost perfect agreement for conventional VI-RADS and "NCE-VI-RADS with dDLR" between the two readers. CONCLUSION: NCE-VI-RADS achieved predictive accuracy for muscle invasion comparable to that of conventional VI-RADS. Additional use of dDLR improved the diagnostic accuracy of NCE-VI-RADS. KEY POINTS: • Non-contrast-enhanced Vesical Imaging Reporting and Data System (NCE-VI-RADS) was developed to avoid risk related to gadolinium-based contrast agent administration. • NCE-VI-RADS had predictive accuracy for muscle invasion comparable to that of conventional VI-RADS. • The additional use of denoising deep learning reconstruction (dDLR) might further improve the diagnostic accuracy of NCE-VI-RADS.

Proposal for a New Vesical Imaging-Reporting and Data System (VI-RADS)-Based Algorithm for the Management of Bladder Cancer: A Paradigm Shift From the Current Transurethral Resection of Bladder Tumor (TURBT)-Dependent Practice.

Transurethral resection of bladder tumor (TURBT) is the essential first step in the current algorithm for the management of bladder cancer (BC). However, despite its necessity and significance, TURBT has several limitations, including cost, hospitalization, anesthesia, potential complications such as bladder perforation, and delay to radical cystectomy. The Vesical Imaging Reporting and Data System (VI-RADS) was developed to standardize the reporting of multiparametric magnetic resonance imaging for BC, and its diagnostic accuracy to predict muscle invasion has been validated. Given the high sensitivity of VI-RADS ≥ 3 and high specificity of VI-RADS ≥ 4 as clinically relevant cutoff values, we herein propose a new VI-RADS-based algorithm for the management of BC. Using this algorithm, patients with VI-RADS ≤ 2 may not need to undergo sampling of the detrusor muscle nor second TURBT even if there is no muscle in the initial TURBT specimen, whereas patients with VI-RADS ≥ 4 may skip conventional TURBT aimed at pathologic confirmation of muscle invasion and immediately undergo radical cystectomy. Our newly proposed algorithm enables the avoidance of unnecessary deep resection or second TURBT as well as delay to radical cystectomy. The VI-RADS-based algorithm enables a paradigm shift from the current TURBT-dependent practice in the management of BC.

Prospective Validation of Vesical Imaging-Reporting and Data System Using a Next-Generation Magnetic Resonance Imaging Scanner-Is Denoising Deep Learning Reconstruction Useful?

PURPOSE: The Vesical Imaging Reporting and Data System (VI-RADS) was launched in 2018 to standardize reporting of magnetic resonance imaging for bladder cancer. This study aimed to prospectively validate VI-RADS using a next-generation magnetic resonance imaging scanner and to investigate the usefulness of denoising deep learning reconstruction. MATERIALS AND METHODS: We prospectively enrolled 98 patients who underwent bladder multiparametric magnetic resonance imaging using a next-generation magnetic resonance imaging scanner before transurethral resection of bladder tumor. Tumors were categorized according to VI-RADS, and we ultimately analyzed 68 patients with pathologically confirmed urothelial bladder cancer. We used receiving operating characteristic curve analyses to assess the predictive accuracy of VI-RADS for muscle invasion. Sensitivity, specificity, positive/negative predictive value, accuracy and area under the curve were calculated for different VI-RADS score cutoffs. RESULTS: Muscle invasion was detected in the transurethral resection of bladder tumor specimens of 18 patients (26%). The optimal cutoff value of the VI-RADS score was determined as ≥4 based on the receiver operating curve analyses. The accuracy of diagnosing muscle invasion using a cutoff of VI-RADS ≥4 was 94% (AUC 0.92). Additionally, we assessed the utility of denoising deep learning reconstruction. Combination with denoising deep learning reconstruction significantly improved the AUC of category by T2-weighted imaging, and of the 4 patients who were misdiagnosed by the final VI-RADS score 3 were correctly diagnosed by T2-weighted imaging+denoising deep learning reconstruction. CONCLUSIONS: In this prospective validation study with a next-generation magnetic resonance imaging scanner, VI-RADS showed high predictive accuracy for muscle invasion in patients with bladder cancer before transurethral resection of bladder tumor. Combining T2-weighted imaging with denoising deep learning reconstruction might further improve the diagnostic accuracy of VI-RADS.

Clinical applications of digital angiography with the harmonization function in body interventional radiology.

Digital subtraction angiography (DSA) is frequently applied in interventional radiology (IR). When DSA is not useful due to misregistration, digital angiography (DA) as an alternative option is used. In DA, the harmonization function (HF) works in real time by harmonizing the distribution of gray steps or reducing the dynamic range; thus, it can compress image gradations, decrease image contrast, and suppress halation artifacts. DA with HF as a good alternative to DSA is clinically advantageous in body IR for generating DSA-like images and simultaneously reducing various motion artifacts and misregistrations caused by patient body motion, poor breath-holding, bowel and ureter peristalsis, and cardiac pulsation as well as halation artifacts often stemming from the lung field. Free-breath DA with HF can improve body IR workflow and decrease the procedure time by reducing the risk of catheter dislocation and using background structures as anatomical landmarks, demonstrating reduced radiation exposure relative to DSA. Thus, HF should be more widely and effectively utilized for appropriate purposes in body IR. This article illustrates the basic facts and principles of HF in DA, and demonstrates clinical advantages and limitations of this function in body IR.