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.

The complexity of bony malalignment in patellofemoral disorders: femoral and tibial torsion, trochlear dysplasia, TT-TG distance, and frontal mechanical axis correlate with each other.

PURPOSE: Several anatomic risk factors associated with patellofemoral disorders have been described. The purpose of this study was to analyze the relationship between bony parameters commonly used to analyze and define patellofemoral malalignment. METHODS: Patients with patellofemoral disorders presenting between 2016 and 2018 who underwent a standardized radiographic workup including conventional radiographs, weight bearing full-leg radiographs, magnetic resonance imaging (MRI) of the knee, and torsional analysis using hip-knee-ankle MRI were initially included. Patients with a history of lower extremity fracture and a history of surgical procedures affecting bony alignment or partial/total arthroplasty were subsequently excluded. Radiographs and MRI of all included patients were analyzed by four independent observers. Parameters of interest were: femoral torsion, tibial torsion, trochlear dysplasia, tibial tuberosity-trochlear groove (TT-TG) distance, and frontal mechanical axis. All parameters were compared between patients with low grade and high grade trochlear dysplasia as well as between female and male patients. Correlation of continuous variables was assessed with the Pearson correlation coefficient. A binary logistic regression model was used for the calculation of odds ratio between different parameters. Interclass correlation coefficients (ICC) were calculated to determine the interobserver reproducibility. RESULTS: A total of 151 patients could be included for detailed analysis. Group comparison revealed that patients with high grade trochlear dysplasia showed significantly higher values for femoral torsion (low grade: 9.8° ± 11.0°, high grade: 16.8° ± 11.5°; p < 0.001) and significantly higher values for TT-TG distance (low grade: 19.0 mm ± 5.0 mm, high grade: 21.9 mm ± 5.4 mm; p = 0.002). No significant difference was found for age, tibial torsion, and frontal mechanical axis. With regard to gender, female patients had higher values for femoral torsion (female: 15.6° ± 11.3°, male: 11.0° ± 12.7°; p = 0.044). The correlation analysis found significant correlation between femoral torsion and tibial torsion (r = 0.244, p = 0.003), femoral torsion and TT-TG distance (r = 0.328, p < 0.001), femoral torsion and frontal mechanical axis (r = 0.291, p < 0.001), and tibial torsion and TT-TG distance (r = 0.182, p = 0.026). CONCLUSION: Bony malalignment in patients with patellofemoral disorder is a complex problem given the significant correlation between femoral and tibial torsion, trochlear dysplasia, TT-TG distance, and frontal mechanical axis. Advanced imaging to analyze rotational and frontal plane alignment is recommended in patients with trochlear dysplasia and/or increased TT-TG on standard radiographs and knee MRI. Understanding of the bony pathology in patellofemoral disorders is key to improve the therapeutic and surgical decision. LEVEL OF EVIDENCE: III, retrospective cohort study.

Correction to: The complexity of bony malalignment in patellofemoral disorders: femoral and tibial torsion, trochlear dysplasia, TT-TG distance, and frontal mechanical axis correlate with each other.

The original article can be found online.

Evaluation of a shortened cardiac MRI protocol for left ventricular examinations: diagnostic performance of T1-mapping and myocardial function analysis.

BACKGROUND: In this study we sought to retrospectively evaluate whether a very brief cardiac magnetic resonance imaging (CMR) protocol sufficiently distinguishes patients with relevant myocardial changes with need for further examination from healthy subjects. METHODS: Patients with clinical indication for CMR (n = 160) were included in the study. Patients were categorized into two groups depending on presence of left ventricular (LV) dysfunction. ROC-analysis was done for results of T1-, T2- mapping and extracellular volume evaluation in patients without LV dysfunction. Binary endpoint was correctly depicted pathology of the conventional qualitative CMR techniques and report. RESULTS: In the patient cohort without LV dysfunction (49%), AUC for T1 mapping was 82% (p < 0.001), 60% for T2 mapping (p = 0.1) and 79% for ECV (p < 0.001). T1 mapping was significantly superior to T2 mapping to rule out left ventricular pathology (p = 0.012). Sensitivity for the combined use of T1 mapping and sBTFE cine imaging was 98%; the negative predictive value was 90%. In 49 patients (30%) full protocol CMR did not provide any additional information; T1 mapping correctly detected 57% of the subjects from this group who would not benefit from additional CMR. CONCLUSION: A shortened CMR protocol comprising T1 mapping and LV-function analysis seems suitable to rule out myocardial alterations. Every third patient of the study population did not benefit from full contrast enhanced CMR. The shortened protocol correctly identified every fifth patient who would not benefit but no relevant pathologic findings with the obligation for treatment were missed.