Algorithmic transparency and interpretability measures improve radiologists' performance in BI-RADS 4 classification.

OBJECTIVE: To evaluate the perception of different types of AI-based assistance and the interaction of radiologists with the algorithm's predictions and certainty measures. METHODS: In this retrospective observer study, four radiologists were asked to classify Breast Imaging-Reporting and Data System 4 (BI-RADS4) lesions (n = 101 benign, n = 99 malignant). The effect of different types of AI-based assistance (occlusion-based interpretability map, classification, and certainty) on the radiologists' performance (sensitivity, specificity, questionnaire) were measured. The influence of the Big Five personality traits was analyzed using the Pearson correlation. RESULTS: Diagnostic accuracy was significantly improved by AI-based assistance (an increase of 2.8% ± 2.3%, 95 %-CI 1.5 to 4.0 %, p = 0.045) and trust in the algorithm was generated primarily by the certainty of the prediction (100% of participants). Different human-AI interactions were observed ranging from nearly no interaction to humanization of the algorithm. High scores in neuroticism were correlated with higher persuasibility (Pearson's r = 0.98, p = 0.02), while higher consciousness and change of accuracy showed an inverse correlation (Pearson's r = -0.96, p = 0.04). CONCLUSION: Trust in the algorithm's performance was mostly dependent on the certainty of the predictions in combination with a plausible heatmap. Human-AI interaction varied widely and was influenced by personality traits. KEY POINTS: • AI-based assistance significantly improved the diagnostic accuracy of radiologists in classifying BI-RADS 4 mammography lesions. • Trust in the algorithm's performance was mostly dependent on the certainty of the prediction in combination with a reasonable heatmap. • Personality traits seem to influence human-AI collaboration. Radiologists with specific personality traits were more likely to change their classification according to the algorithm's prediction than others.

90Y Radioembolization in the Treatment of Neuroendocrine Neoplasms: Results of an International Multicenter Retrospective Study.

In neuroendocrine neoplasms (NENs), the presence of distant metastases has a severe impact on survival leading to a relevant decrease in the 5-y survival rate. Here, 90Y radioembolization (90Y RE) might be an important treatment option; however, data to support clinical benefits for 90Y RE are scarce. Therefore, the purpose of this study was to analyze the use of 90Y RE in NEN patients with hepatic metastases in an international, multicenter retrospective analysis and assess the potential role of 90Y RE in a multimodal treatment concept. Methods: In total, 297 angiographic evaluations in NEN patients before 90Y RE were analyzed. Baseline characteristics and parameters derived from imaging evaluation and 90Y RE were analyzed. Tumor response was assessed using RECIST 1.1, and survival data were collected. Mean overall survival (OS) between different groups was compared using Kaplan-Meier curves and the log rank test. A P value of less than 0.05 indicated statistical significance. Results: After 90Y RE, the disease control rate according to RECIST 1.1 was 83.5% after 3 mo and 50.9% after 12 mo. OS in the entire population was 38.9 ± 33.0 mo. High tumor grade (P < 0.006) and high tumor burden (P = 0.001) were both associated with a significant decrease in OS. The presence of extrahepatic metastases (P = 0.335) and the type of metastatic vascularization pattern (P = 0.460) had no influence on OS. Patients who received 90Y RE as second-line therapy had a slightly longer but not statistically significant OS than patients who had 90Y RE in a salvage setting (44.8 vs. 30.6 mo, P = 0.078). Hepatic and global progression-free survival after 90Y RE was significantly decreased in heavily pretreated patients, compared with patients with second-line therapy (P = 0.011 and P = 0.010, respectively). Conclusion:90Y RE could be an important alternative to peptide receptor radionuclide therapy as second-line treatment in patients with progressive liver-dominant disease pretreated with somatostatin analogs.

3D grating-based X-ray phase-contrast computed tomography for high-resolution quantitative assessment of cartilage: An experimental feasibility study with 3T MRI, 7T MRI and biomechanical correlation.

OBJECTIVE: Aim of this study was, to demonstrate the feasibility of high-resolution grating-based X-ray phase-contrast computed tomography (PCCT) for quantitative assessment of cartilage. MATERIALS AND METHODS: In an experimental setup, 12 osteochondral samples were harvested from n = 6 bovine knees (n = 2 each). From each knee, one cartilage sample was degraded using 2.5% Trypsin. In addition to PCCT and biomechanical cartilage stiffness measurements, 3T and 7T MRI was performed including MSME SE T2 and ME GE T2* mapping sequences for relaxationtime measurements. Paired t-tests and receiver operating characteristics (ROC) curves were used for statistical analyses. RESULTS: PCCT provided high-resolution images for improved morphological cartilage evaluation as compared to 3T and 7T MRI. Quantitative analyses revealed significant differences between the superficial and the deep cartilage layer for T2 mapping as well as for PCCT (P<0.05). No significant difference was detected for PCCT between healthy and degraded samples (P>0.05). MRI and stiffness measurements showed significant differences between healthy and degraded osteochondral samples. Accuracy in the prediction of cartilage degradation was excellent for MRI and biomechanical analyses. CONCLUSION: In conclusion, high-resolution grating-based X-ray PCCT cartilage imaging is feasible. In addition to MRI and biomechanical analyses it provides complementary, water content independent, information for improved morphological and quantitative characterization of articular cartilage ultrastructure.

Cathepsin S is associated with degradation of collagen I in abdominal aortic aneurysm.

BACKGROUND: Cathepsins have been described in the pathogenesis of abdominal aortic aneurysm (AAA), their exact role, especially in collagen degradation, is still unclear. The aim of the present study was therefore to analyse relevant cathepsins in human AAA tissue samples in relation to collagen I, III, and their degradation products. MATERIALS AND METHODS: Samples from 37 AAA patients obtained from elective open surgical repair and eight healthy non-aneurysmatic aortas from kidney donors were included. Expression of cathepsins B, D, K, L, S, cystatin C, collagen I and III, their degraded products C-Telopeptide of type 1 and 3 collagen (CTX-I, CTX-III), cellular markers for leukocytes (CD45), T cells (CD3), macrophage scavenger receptor-1 (MSR-1), synthetic, and contractile smooth muscle cells (SMCs) (smoothelin: SMTH, collagen I and III, myosin heavy chain: MHC, embryonic smooth muscle myosin heavy chain: SMemb) were determined at messenger RNA (mRNA) level, using SYBRGreen-based quantitative PCR and at protein level using enzyme-linked immunosorbent assay (ELISA). RESULTS: Expression of cathepsins B, D, L, and S at mRNA level was significantly elevated in AAA compared to control aorta (1.7-fold, p = 0.025; 2.5-fold, p = 0.002; 2.6-fold, p = 0.034; and 7.0-fold, p = 0.003). Expression of cathepsin S correlated significantly with leukocytes and macrophages (ρ = 0.398, p = 0.033 and ρ = 0.422, p = 0.020), synthetic SMCs were significantly associated with cathepsins B, D, and L (ρ = 0.522, p = 0.003; ρ = 0.431, p = 0.015 and ρ = 0.467, p = 0.008). At protein level, cathepsins B and S were elevated in AAA compared to controls (5.4-fold, p = 0.001 and 7.3-fold, p < 0.001). Significant correlations were observed between collagen I, its degraded product, and cathepsin S (r = -0.350, p = 0.034 and r = +0.504, p < 0.001). Expression of cathepsin B was associated with SMCs, expression of cathepsin S with inflammatory cells. CONCLUSIONS: Particularly cathepsin S was associated with the degradation product of collagen I and thus might be involved in the progression of AAA. Furthermore, cathepsin S correlated with inflammatory cells.

Quantitative expression and localization of cysteine and aspartic proteases in human abdominal aortic aneurysms.

Cysteine and aspartic proteases possess high elastolytic activity and might contribute to the degradation of the abdominal aortic aneurysm (AAA) wall. The aim of this study was to analyze, in detail, the proteases (cathepsins B, D, K, L and S, and inhibitor cystatin C) found in human AAA and healthy aortic tissue samples. The vessel walls from AAA patients (n=36) and nonaneurysmal aortae (n=10) were retrieved using conventional surgical repair and autopsy methods. Serum samples from the same AAA patients and 10 healthy volunteers were also collected. Quantitative expression analyses were performed at the mRNA level using real-time reverse transcriptase-PCR (RT-PCR). Furthermore, analyses at the protein level included western blot and immunoprecipitation analyses. Cellular sources of cysteine/aspartic proteases and cystatin C were identified by immunohistochemistry (IHC). All cysteine/aspartic proteases and cystatin C were detected in the AAA and control samples. Using quantitative RT-PCR, a significant increase in expression was observed for cathepsins B (P=0.021) and L (P=0.018), compared with the controls. Cathepsin B and cystatin C were also detected in the serum of AAA patients. Using IHC, smooth muscle cells (SMCs) and macrophages were positive for all of the tested cathepsins, as well as cystatin C; in addition, the lymphocytes were mainly positive for cathepsin B, followed by cathepsins D and S. All cysteine/aspartic proteases analyzed in our study were detected in the AAA and healthy aorta. The highest expression was found in macrophages and SMCs. Consequently, cysteine/aspartic proteases might play a substantial role in AAA.

Histopathological analysis of cellular localization of cathepsins in abdominal aortic aneurysm wall.

An important feature of abdominal aortic aneurysm (AAA) is the destruction of vessel wall, especially elastin and collagen. Besides matrix metalloproteinases, cathepsins are the most potent elastolytic enzymes. The expression of cathepsins with known elastolytic and collagenolytic activities in the individual cells within AAA has not yet been determined. The vessel wall of 32 AAA patients and 10 organ donors was analysed by immunohistochemistry for expression of cathepsins B, D, K, L and S, and cystatin C in all cells localized within AAA. Luminal endothelial cells (ECs) of AAA were positive for cathepsin D and partially for cathepsins B, K and S. Endothelial cells of the neovessels and smooth muscle cells in the media were positive for all cathepsins tested, especially for cathepsin B. In the inflammatory infiltrate all cathepsins were expressed in the following pattern: B > D = S > K = L. Macrophages showed the highest staining intensity for all cathepsins. Furthermore, weak overall expression of cystatin C was observed in all the cells localized in the AAA with the exception of the ECs. There is markedly increased expression of the various cathepsins within the AAA wall compared to healthy aorta. Our data are broadly consistent with a role for cathepsins in AAA; and demonstrate expression of cathepsins D, B and S in phagocytic cells in the inflammatory infiltrate; and also may reveal a role for cathepsin B in lymphocytes.

Expression of a disintegrin and metalloprotease in human abdominal aortic aneurysms.

OBJECTIVES: The a disintegrin and metalloprotease (ADAM) family of metalloproteases possesses a proteolytic function and activates various inflammatory factors. Their expression pattern in an abdominal aortic aneurysm (AAA) is as yet unknown. The aim of this study was to make a detailed analysis of the expression of ADAMs 8, 9, 10, 12, 15 and 17, and their tissue inhibitors of metalloprotease (TIMP)-1 and TIMP-3 in patients with AAA. DESIGN: The aortic vessel walls of AAA patients (n=20) and non-aneurysmal aortic specimens (n=10) were obtained by conventional surgical repair and autopsy. SYBR green-based real-time PCR, histology and immunohistochemistry were performed on all samples. MAIN OUTCOME MEASURES: Quantitative expression analysis and the localisation of various ADAMs in AAA. RESULTS: ADAMs tested in our study were expressed in both AAA and control aorta without any significant differences between the groups. In contrast, expression of TIMP-1 was significantly reduced in AAA compared to control vessels. Smooth muscle cells (SMCs), neovessels and macrophages were positive for all ADAMs and TIMPs tested. Infiltrates were negative for TIMP-3, and luminal endothelial cells were positive for ADAMs 15 and 17. A significant positive correlation was observed between ADAMs 10, 12, 15, 17, TIMP-3 and SMCs. CONCLUSION: ADAMs are constitutively expressed in normal aortic vessel walls and AAA, particularly in SMCs.