Deep learning-based k-space-to-image reconstruction and super resolution for diffusion-weighted imaging in whole-spine MRI.

PURPOSE: To assess the feasibility of deep learning (DL)-based k-space-to-image reconstruction and super resolution for whole-spine diffusion-weighted imaging (DWI). METHOD: This retrospective study included 97 consecutive patients with hematologic and/or oncologic diseases who underwent DL-processed whole-spine MRI from July 2022 to March 2023. For each patient, conventional (CONV) axial single-shot echo-planar DWI (b = 50, 800 s/mm2) was performed, followed by DL reconstruction and super resolution processing. The presence of malignant lesions and qualitative (overall image quality and diagnostic confidence) and quantitative (nonuniformity [NU], lesion contrast, signal-to-noise ratio [SNR], contrast-to-noise ratio [CNR], and ADC values) parameters were assessed for DL and CONV DWI. RESULTS: Ultimately, 67 patients (mean age, 63.0 years; 35 females) were analyzed. The proportions of vertebrae with malignant lesions for both protocols were not significantly different (P: [0.55-0.99]). The overall image quality and diagnostic confidence scores were higher for DL DWI (all P ≤ 0.002) than CONV DWI. The NU, lesion contrast, SNR, and CNR of each vertebral segment (P ≤ 0.04) but not the NU of the sacral segment (P = 0.51) showed significant differences between protocols. For DL DWI, the NU was lower, and lesion contrast, SNR, and CNR were higher than those of CONV DWI (median values of all segments; 19.8 vs. 22.2, 5.4 vs. 4.3, 7.3 vs. 5.5, and 0.8 vs. 0.7). Mean ADC values of the lesions did not significantly differ between the protocols (P: [0.16-0.89]). CONCLUSIONS: DL reconstruction can improve the image quality of whole-spine diffusion imaging.

Differentiation of multiple myelomas from osteolytic bone metastases: Diagnostic value of tumor homogeneity on Contrast-Enhanced CT.

OBJECTIVES: To investigate the diagnostic value of tumor homogeneity on contrast-enhanced (CE) computed tomography (CT) to differentiate multiple myeloma (MM) from osteolytic bone metastases (Mets). METHODS: This retrospective study included patients who were diagnosed with MM or Mets and had multiple (≥2) osteolytic bone tumors on pre-treatment CE-CT. Intratumoral homogeneity was assessed by coefficient of variation (CV, ratio of standard deviation to mean) of the density of a single lesion (CV-lesion). Intertumoral homogeneity was assessed as the CV of the densities of multiple lesions in one patient (CV-patient). A classification model was built from CT parameters using classification and regression tree (CART) analysis. Diagnostic performance of the model was evaluated using C-statistics. RESULTS: A total of 272 lesions (81 MM and 191 Mets) of 105 patients were analyzed. The mean CV-lesion and CV-patient of MM were significantly lower than those of Mets: 0.17 vs 0.26 for CV-lesion (p = 0.005) and 0.16 vs 0.23 for CV-patient (p = 0.013). Thickened struts were more common in MM than in Mets (49.1% vs 12.8%, p ≤ 0.001). In CART analysis, CV-lesion was the first partitioning predictor, followed by thickened struts and by CV patient. The CART model could distinguish MM from Mets in both the model development cohort (C-statistic: 0.843) and the temporal validation cohort (0.721, 0.686, and 0.686 for three reviewers, respectively). CONCLUSIONS: MM showed intratumoral and intertumoral homogeneity compared with Mets on CE-CT. The combination of CV-lesion and CV-patient can be helpful to radiologists in differentiation of MM from Mets. ADVANCES IN KNOWLEDGE: Our study showed that MM had intratumoral and intertumoral homogeneity compared with Mets on contrast-enhanced CT.

Per-Feature Accuracy of Liver Imaging Reporting and Data System Locoregional Treatment Response Algorithm: A Systematic Review and Meta-Analysis.

We aimed to investigate the accuracy of each imaging feature of LI-RADS treatment response (LR-TR) viable category for diagnosing tumor viability of locoregional therapy (LRT)-treated HCC. Studies evaluating the per feature accuracy of the LR-TR viable category on dynamic contrast-enhanced CT or MRI were identified in databases. A bivariate random-effects model was used to calculate the pooled sensitivity, specificity, and diagnostic odds ratio (DOR) of LR-TR viable features. Ten studies assessing the accuracies of LR-TR viable features (1153 treated observations in 971 patients) were included. The pooled sensitivities and specificities for diagnosing viable HCC were 81% (95% confidence interval [CI], 63-92%) and 95% (95% CI, 88-98%) for nodular, mass-like, or irregular thick tissue (NMLIT) with arterial phase hyperenhancement (APHE), 55% (95% CI, 34-75%) and 96% (95% CI, 94-98%) for NMLIT with washout appearance, and 21% (95% CI, 6-53%) and 98% (95% CI, 92-100%) for NMLIT with enhancement similar to pretreatment, respectively. Of these features, APHE showed the highest pooled DOR (81 [95% CI, 25-261]), followed by washout appearance (32 [95% CI, 13-82]) and enhancement similar to pretreatment (14 [95% CI, 5-39]). In conclusion, APHE provided the highest sensitivity and DOR for diagnosing viable HCC following LRT, while enhancement similar to pretreatment showed suboptimal performance.