Automated segmentation of brain metastases with deep learning: a multi-center, randomized crossover, multi-reader evaluation study.

BACKGROUND: Artificial intelligence has been proposed for brain metastasis (BM) segmentation but it has not been fully clinically validated. The aim of this study was to develop and evaluate a system for BM segmentation. METHODS: A deep-learning-based BM segmentation system (BMSS) was developed using contrast-enhanced MR images from 488 patients with 10,338 brain metastases. A randomized crossover, multi-reader study was then conducted to evaluate the performance of the BMSS for BM segmentation using data prospectively collected from 50 patients with 203 metastases at five centers. Five radiology residents and five attending radiologists were randomly assigned to contour the same prospective set in assisted and unassisted modes. Aided and unaided Dice similarity coefficients (DSCs) and contouring times per lesion were compared. RESULTS: The BMSS alone yielded a median DSC of 0.91 (95% confidence interval, 0.90-0.92) in the multi-center set and showed comparable performance between the internal and external sets (p = 0.67). With BMSS assistance, the readers increased the median DSC from 0.87 (0.87-0.88) to 0.92 (0.92-0.92) (p < 0.001) with a median time saving of 42% (40-45%) per lesion. Resident readers showed a greater improvement than attending readers in contouring accuracy (improved median DSC, 0.05 [0.05-0.05] vs. 0.03 [0.03-0.03]; p < 0.001), but a similar time reduction (reduced median time, 44% [40-47%] vs. 40% [37-44%]; p = 0.92) with BMSS assistance. CONCLUSIONS: The BMSS can be optimally applied to improve the efficiency of brain metastasis delineation in clinical practice.

A Modified Targetoid Feature Emphasizing Thin-Rim APHE to Improve the Diagnostic Performance of LI-RADS for Malignant Hepatic Tumors.

Objective: To identify imaging features that help distinguish between HCCs and non-HCC malignancies assigned to LI-RADS M (LR-M) and evaluate the diagnostic performance of a LI-RADS with targetoid criteria using thin-rim arterial phase hyperenhancement (APHE). Materials and Methods: This retrospective study included 381 patients (387 observations) at high-risk for HCC who underwent enhanced-MRI before surgery. Three radiologists reviewed images for LI-RADS categorization of hepatic observations. Univariate and multivariate analysis was conducted to determine reliable features to differentiate between HCC and non-HCC malignancies among the LR-M lesions. The thin-rim (<30%) APHE was defined based on the thickest thickness of rim APHE compared with the tumor radius, and a modified LI-RADS emphasizing thin-rim APHE as a specific feature of LR-M was established. We compared the diagnostic performance of modified LR-M and LI-RADS 5 (LR-5) with the conventional one. Results: Thin-rim APHE and targetoid diffusion-weighted imaging (DWI) were found as independent predictive factors of non-HCC malignancies, while enhancing capsule, thick-rim APHE and peripheral washout were noted as independent variables significantly associated with HCC of LR-M (P<0.05). The noticeable diagnostic performance of thin-rim APHE in distinguishing non-HCC malignancies from HCCs using the ROC curve. Emphasizing thin-rim APHE on targetoid features, the modified LR-M revealed significantly superior specificity and accuracy (89.4% vs 81.1%, P=0.004; and 87.9% vs 82.2%, P=0.027, respectively) while maintaining high sensitivity (82.2% vs 86.0%; P=0.529) compared with the LR-M. Meanwhile, the modified LR-5 achieved greater sensitivity and accuracy (88.6% vs 79.7%, P=0.004; and 85.8% vs 80.1%, P=0.036, respectively) for diagnosing HCC, without compromising specificity (78.3% vs.81.1%; P=0.608) compared with the LR-5. Conclusion: Thin-rim APHE may be the specific imaging feature for differentiating non-HCC malignancies from HCCs within LR-M. The modified targetoid criteria emphasizing thin-rim APHE can improve the diagnostic performance of LI-RADS for hepatic malignancies.

AI-assisted compressed sensing and parallel imaging sequences for MRI of patients with nasopharyngeal carcinoma: comparison of their capabilities in terms of examination time and image quality.

OBJECTIVE: To compare examination time and image quality between artificial intelligence (AI)-assisted compressed sensing (ACS) technique and parallel imaging (PI) technique in MRI of patients with nasopharyngeal carcinoma (NPC). METHODS: Sixty-six patients with pathologically confirmed NPC underwent nasopharynx and neck examination using a 3.0-T MRI system. Transverse T2-weighted fast spin-echo (FSE) sequence, transverse T1-weighted FSE sequence, post-contrast transverse T1-weighted FSE sequence, and post-contrast coronal T1-weighted FSE were obtained by both ACS and PI techniques, respectively. The signal-to-noise ratio (SNR), contrast-to-noise ratio (CNR), and duration of scanning of both sets of images analyzed by ACS and PI techniques were compared. The images from the ACS and PI techniques were scored for lesion detection, margin sharpness of lesions, artifacts, and overall image quality using the 5-point Likert scale. RESULTS: The examination time with ACS technique was significantly shorter than that with PI technique (p < 0.0001). The comparison of SNR and CNR showed that ACS technique was significantly superior with PI technique (p < 0.005). Qualitative image analysis showed that the scores of lesion detection, margin sharpness of lesions, artifacts, and overall image quality were higher in the ACS sequences than those in the PI sequences (p < 0.0001). Inter-observer agreement was evaluated for all qualitative indicators for each method, in which the results showed satisfactory-to-excellent agreement (p < 0.0001). CONCLUSION: Compared with the PI technique, the ACS technique for MR examination of NPC can not only shorten scanning time but also improve image quality. CLINICAL RELEVANCE STATEMENT: The artificial intelligence (AI)-assisted compressed sensing (ACS) technique shortens examination time for patients with nasopharyngeal carcinoma, while improving the image quality and examination success rate, which will benefit more patients. KEY POINTS: • Compared with the parallel imaging (PI) technique, the artificial intelligence (AI)-assisted compressed sensing (ACS) technique not only reduced examination time, but also improved image quality. • Artificial intelligence (AI)-assisted compressed sensing (ACS) pulls the state-of-the-art deep learning technique into the reconstruction procedure and helps find an optimal balance of imaging speed and image quality.

Response evaluation of hepatocellular carcinoma treated with stereotactic body radiation therapy: magnetic resonance imaging findings.

PURPOSE: To summarize the magnetic resonance imaging manifestations of hepatocellular carcinoma (HCC) with and without progression after stereotactic body radiation therapy (SBRT) and evaluate the treatment effect using the modified Liver Reporting and Data System (LI-RADS). METHODS: Between January 2015 and December 2020, 102 patients with SBRT-treated HCC were included. Tumor size, signal intensity, and enhancement patterns at each follow-up period were analyzed. Three different patterns of enhancement: APHE and wash-out, non-enhancement, and delayed enhancement. For modified LI-RADS, delayed enhancement with no size increase were considered to be a "treatment-specific expected enhancement pattern" for LR-TR non-viable. RESULTS: Patients were divided into two groups: without (n = 96) and with local progression (n = 6). Among patients without local progression, APHE and wash-out pattern demonstrated conversion to the delayed enhancement (71.9%) and non-enhancement (20.8%) patterns, with decreased signal intensity on T1WI(92.9%) and DWI(99%), increased signal intensity on T1WI (99%), and decreased size. The signal intensity and enhancement patterns stabilized after 6-9 months. Six cases with progression exhibited tumor growth, APHE and wash-out, and increased signal intensity on T2WI/DWI. Based on the modified LI-RADS criteria, 74% and 95% showed LR-TR-nonviable in 3 and 12 months post-SBRT, respectively. CONCLUSIONS: After SBRT, the signal intensity and enhancement patterns of HCCs showed a temporal evolution. Tumor growth, APHE and wash-out, and increased signal intensity on T2WI/DWI indicates tumor progression. Modified LI-RADS criteria showed good performance in evaluating nonviable lesions after SBRT.

False-negative and false-positive outcomes of computer-aided detection on brain metastasis: Secondary analysis of a multicenter, multireader study.

BACKGROUND: Errors have seldom been evaluated in computer-aided detection on brain metastases. This study aimed to analyze false negatives (FNs) and false positives (FPs) generated by a brain metastasis detection system (BMDS) and by readers. METHODS: A deep learning-based BMDS was developed and prospectively validated in a multicenter, multireader study. Ad hoc secondary analysis was restricted to the prospective participants (148 with 1,066 brain metastases and 152 normal controls). Three trainees and 3 experienced radiologists read the MRI images without and with the BMDS. The number of FNs and FPs per patient, jackknife alternative free-response receiver operating characteristic figure of merit (FOM), and lesion features associated with FNs were analyzed for the BMDS and readers using binary logistic regression. RESULTS: The FNs, FPs, and the FOM of the stand-alone BMDS were 0.49, 0.38, and 0.97, respectively. Compared with independent reading, BMDS-assisted reading generated 79% fewer FNs (1.98 vs 0.42, P < .001); 41% more FPs (0.17 vs 0.24, P < .001) but 125% more FPs for trainees (P < .001); and higher FOM (0.87 vs 0.98, P < .001). Lesions with small size, greater number, irregular shape, lower signal intensity, and located on nonbrain surface were associated with FNs for readers. Small, irregular, and necrotic lesions were more frequently found in FNs for BMDS. The FPs mainly resulted from small blood vessels for the BMDS and the readers. CONCLUSIONS: Despite the improvement in detection performance, attention should be paid to FPs and small lesions with lower enhancement for radiologists, especially for less-experienced radiologists.

Discriminating rectal cancer grades using restriction spectrum imaging.

PURPOSE: Restriction spectrum imaging (RSI) is a novel diffusion MRI model that separates water diffusion into several microscopic compartments. The restricted compartment correlating to the tumor cellularity is expected to be a potential indicator of rectal cancer aggressiveness. Our aim was to assess the ability of RSI model for rectal tumor grading. METHODS: Fifty-eight patients with different rectal cancer grading confirmed by biopsy were involved in this study. DWI acquisitions were performed using single-shot echo-planar imaging (SS-EPI) with multi-b-values at 3 T. We applied a three-compartment RSI model, along with ADC model and diffusion kurtosis imaging (DKI) model, to DWI images of 58 patients. ROC and AUC were used to compare the performance of the three models in differentiating the low grade (G1 + G2) and high grade (G3). Mean ± standard deviation, ANOVA, ROC analysis, and correlation analysis were used in this study. RESULTS: The volume fraction of restricted compartment C1 from RSI was significantly correlated with grades (r = 0.403, P = 0.002). It showed significant difference between G1 and G3 (P = 0.008) and between G2 and G3 (P = 0.01). As for the low-grade and high-grade discrimination, significant difference was found in C1 (P < 0.001). The AUC of C1 for differentiation between low-grade and high-grade groups was 0.753 with a sensitivity of 72.0% and a specificity of 70.0%. CONCLUSION: The three-compartment RSI model was able to discriminate the rectal cancer of low and high grades. The results outperform the traditional ADC model and DKI model in rectal cancer grading.

Radiological-pathological correlation of yolk sac tumor in 20 patients.

BACKGROUND: Yolk sac tumor (YST) is a rare tumor. Familiarity of its radiological characteristics may permit preoperative diagnosis and improve surgical management of patients. However, a detailed description of the imaging features of YST with pathological correlation in particular is scarce. PURPOSE: To investigate computed tomography (CT) findings of YSTs with pathological correlation. MATERIAL AND METHODS: CT images of 20 patients with pathologically proven YST were retrospectively reviewed. The location, size, margin, internal architecture, and pattern and degree enhancement of the lesion were evaluated. Radiological findings were correlated with pathological results. RESULTS: The locations of 20 tumors were distributed between the testis (n = 3), ovary (n = 6), sacrococcygeal area (n = 6), rectum (n = 1), and mediastinum (n = 4). The median age was 13 years. On CT images, all tumors were seen as oval (n = 14) or irregular (n = 6), well-defined (n = 16) or ill-defined (n = 4) masses with a mean size of 9.7 cm. The lesions were solid cystic (n = 10), entirely solid (n = 6), or predominantly cystic (n = 4). Intratumoral hemorrhage, calcification, and fatty tissue were seen in nine, three, and two tumors, respectively. Discontinuity of the tumor wall was seen in eight tumors. After contrast media administration, most tumors showed heterogeneous moderate to marked enhancement (n = 7) or heterogeneous marked enhancement (n = 9). Enlarged intratumoral vessels were seen in 17 tumors. CONCLUSION: YST usually appears as a large solid-cystic mass with intratumoral hemorrhage, capsular tear, marked heterogeneous enhancement, and enlarged intratumoral vessels on CT images. Intratumoral calcification and fatty tissue, although rare, may indicate a mixed YST containing teratoma component.

CT imaging of ovarian yolk sac tumor with emphasis on differential diagnosis.

Ovarian yolk sac tumors (YSTs) are rare neoplasms. No radiological study has been done to compare the imaging findings between this type of tumor and other ovarian tumors. Here we analyzed the CT findings of 11 pathologically proven ovarian YSTs and compared their imaging findings with 18 other types of ovarian tumors in the same age range. Patient age, tumor size, tumor shape, ascites and metastasis of two groups did not differ significantly (P > 0.05). A mixed solid-cystic nature, intratumoral hemorrhage, marked enhancement and dilated intratumoral vessel of two groups differed significantly (P < 0.05). The area under the ROC curve of four significant CT features was 0.679, 0.707, 0.705, and 1.000, respectively. Multivariate logistic regression analysis identified two independent signs of YST: intratumoral hemorrhage and marked enhancement. Our results show that certain suggestive CT signs that may be valuable for improving the accuracy of imaging diagnosis of YST and may be helpful in distinguishing YST from other ovarian tumors.

Computed tomography imaging of anterior and middle mediastinal Ewing sarcoma/primitive neuroectodermal tumors.

PURPOSE: We describe the computed tomography (CT) imaging features of Ewing sarcoma (EWS)/primitive neuroectodermal tumors (PNETs) arising in the anterior and middle mediastinum. MATERIALS AND METHODS: The CT imaging findings of 6 cases of anterior and middle mediastinal EWS/PNETs were reviewed retrospectively. All 6 patients were examined with chest radiographs and CT, and 4 patients underwent isotope bone scans. RESULTS: The average patient age was 40 years. Results using unenhanced CT showed lobulated, heterogeneous masses with patchy, necrotic foci in 5 cases, and one small, oval homogenous mass in the sixth case. There was no calcification in any of the cases. The contrast-enhanced CT results demonstrated that there were 4 cases of heterogeneous enhancement and one case of homogeneous enhancement. All the masses were ill-defined, and in 4 cases, the masses were displaced and encompassed the adjacent great vessels. The tumors directly infiltrated the anterior chest wall in 3 cases, and in one of these cases had eroded the sternum. Four cases demonstrated pleural effusions. Isotope bone scans showed distant bone metastases at diagnosis in 2 cases. CONCLUSIONS: EWS/PNETs in the anterior and middle mediastinum appear as ill-defined, heterogenerous masses that are not distinguishable from other, more common, causes of mediastinal masses, based on their CT features.