Insights into Imaging最新文献

Objectives: Artificial intelligence (AI) has gained increasing interest in supporting clinicians in patient selection, treatment planning, and prognostics. However, the current application of AI in the treatment of colorectal liver metastases (CRLM) is not clearly established and validated. This scoping review assesses the current impact of AI techniques on local therapeutic strategies for CRLM, exploring its benefits, challenges, and future directions.

Materials and methods: A comprehensive literature search in PubMed, EMBASE, Web of Science, and SCOPUS was conducted for patients with CRLM undergoing surgery or thermal ablation, along with descriptions of AI tools. Eligible studies were cohort studies or clinical trials. Data extraction focused on treatment strategies, AI techniques, clinical and radiomics parameters, and outcomes.

Results: Out of 1464 articles, thirteen met the inclusion criteria. Eight articles regarded thermal liver ablation, and five surgical resection for CRLM. Most studies used traditional machine learning methods, such as support vector machines and random forests, combined with radiomics for predictive model building. While most studies demonstrated high performance, they frequently involved small sample sizes, and machine learning techniques often lacked robustness.

Conclusions: AI shows promising results in improving local treatment strategies for CRLM, but further advancements are required for AI decision support tools. Future research should focus on large multicentre studies to validate AI-driven personalised colorectal liver metastases treatment strategies.

Critical relevance statement: This review evaluates the use of AI in colorectal liver metastases treatment for outcome prediction and treatment evaluation. Included studies used AI for segmentation and predictive modelling and were often limited by small sample-sized single-centre studies, necessitating multicentre studies.

Key points: The position of AI in local colorectal liver metastases treatments is not yet established. AI algorithms are mostly used for predictive model building and image registration. Studies often lack validity, limiting generalisability and implementation of AI support tools.

Objectives: Except for a few specific lesions, the analysis of the margins of focal bone lesions with MRI has been largely overlooked in the literature. We observed that some lesions exhibited concave margins, suggesting a non-aggressive nature. This study aimed to determine whether concave borders are more frequently present in certain types of focal lesions of the long bones on MRI, particularly in benign lesions.

Materials and methods: MRI examinations of 586 focal intraosseous lesions of the long bones were retrospectively reviewed. Each lesion margin was independently classified by two musculoskeletal radiologists as concave, non-concave, or cortical. The number of concave margins was analyzed according to the lesion type and its classification as tumoral (benign or malignant) or non-tumoral.

Results: The study group included 15 different types of lesions (75.1% tumors, 24.9% non-tumoral lesions). The number of concave margins per lesion varied significantly by lesion type (p < 0.001). Benign lesions and benign tumors had more concave margins than malignant lesions (p < 0.001). All osteonecrosis lesions, 44.8% of fibrous dysplasia lesions, 20.8% of abscesses, 14.8% of chondroblastomas, 12.5% of Langerhans cell histiocytosis lesions, and 2.9% of metastases showed at least two concave margins. No lesions in the other groups had more than one concave margin.

Conclusion: Our study revealed that several intraosseous lesions tend to exhibit concave margins. The presence of at least two smooth and regular inward-curved margins was most commonly found in benign lesions and benign tumors, such as fibrous dysplasia.

Critical relevance statement: This study shows that at least two smooth, inward-curved margins are more common in benign bone lesions and tumors and may help in recognizing fibrous dysplasia in non-fatty lesions.

Key points: Identifying concave margins may help in recognizing benign bone lesions and tumors. Concave margins are mainly observed in osteonecrosis and fibrous dysplasia. Concave margins suggest a non-aggressive nature of focal bone lesions.

Objectives: The notch of Harty (NOH) is a pseudolesion located at the distal tibial plafond that can mimic a clinically significant osteochondral lesion. Reported prevalence in an adult population ranged between 25% and 45%, but the prevalence in younger age groups is unknown. In this study, we assessed the age-dependent prevalence of the NOH in a young population to investigate potential remodeling over time.

Materials and methods: A retrospective analysis of a single-center database included 895 ankle MRIs of patients aged 6-25 years between 2018 and 2022. Cases with significant artifacts or pathological conditions were excluded. The NOH was categorized into types 1 (fluid-filled) and 2 (cartilage-filled), and measurements were taken for mediolateral, craniocaudal, and anteroposterior dimensions. Logistic and linear regression models with restricted cubic splines assessed age-related trends. Radiographic sensitivity for NOH detection was evaluated in 252 radiographs with MRI-confirmed NOH as a secondary endpoint.

Results: The overall prevalence of NOH in this study population was 47.1% (14.9% type 1 and 31.6% type 2). Type 1 prevalence peaked at age 8 and declined into adulthood, while type 2 prevalence peaked at age 14 and stabilized through adolescence. The NOH size increased with age until 16 years and subsequently decreased. Radiographic sensitivity for detecting NOH was 51.2%.

Conclusion: The NOH is more common in younger individuals, with prevalence and size decreasing with age. This supports its classification as a developmental variant characterized by age-related remodeling.

Critical relevance statement: This study provides novel insights into the age-dependent prevalence and remodeling of the NOH, enhancing diagnostic precision and improving differentiation between developmental variants and pathological findings in clinical radiology.

Key points: The study investigates the age-dependent prevalence of the NOH. Results show developmental remodeling patterns with prevalence peaking in younger individuals. Data support the theory of NOH being a developmental variant.

Objectives: Differentiating between nontuberculous mycobacteria (NTM) pulmonary disease (NTM-PD) and colonization (NTM-PC) is clinically important but difficult. It remains unknown whether artificial intelligence utilizing clinical data and chest CT images could address this clinical problem.

Materials and methods: Patients were retrospectively recruited with NTM isolation from respiratory specimens in two hospitals. Their disease or colonization status was determined by three NTM experts. We developed a multimodal deep learning model named NTMNet, which integrates chest CT scans and clinical data (including age, sex, acid-fast smear [AFS] results, and mycobacterial species) to predict NTM disease status. The performance of NTMNet was evaluated on both internal and external test sets.

Results: A total of 324 NTM-PC patients and 285 NTM-PD patients were included. Among the internal and external test sets, the area under the receiver operating characteristic curve (AUC) for predicting NTM disease status using CT imaging was 0.73 (95% CI: 0.62-0.82) and 0.78 (95% CI: 0.75-0.83), respectively. When imaging data were integrated with clinical information, our NTMNet model achieved AUC values of 0.85 (95% CI: 0.80-0.93) and 0.82 (95% CI: 0.78-0.89), respectively. Furthermore, our NTMNet model demonstrated comparable accuracy to that of three experienced pulmonologists in determining NTM disease status in the reader study.

Conclusion: Our multimodal NTMNet exhibited satisfactory performance in distinguishing disease status among patients with respiratory NTM isolates. This deep learning-based model has the potential to assist physicians in clinical management, achieving diagnostic accuracy comparable to that of pulmonologists.

Critical relevance statement: A deep learning model leveraging chest computed tomography images and clinical data effectively differentiated NTM disease status, achieving a classification accuracy comparable to that of pulmonologists and demonstrating its potential to support accurate NTM diagnosis in clinical settings.

Key points: Accurately distinguishing nontuberculous mycobacteria (NTM) disease status is clinically important but challenging. The NTMNet model effectively differentiated the NTM disease status and matched the performance of the pulmonologists. The NTMNet model could be a potential diagnostic tool for patients with respiratory NTM isolates.

Objectives: To evaluate the diagnostic value of four-dimensional computed tomography (4DCT) in conjunction with ultrasonography (US) and sestamibi scintigraphy (SeS) with ^99mTc-MIBI, compared to US and SeS alone in primary hyperparathyroidism.

Materials and methods: A retrospective study was conducted on patients who underwent 4DCT for suspected parathyroid adenomas between January 2019 and December 2023, followed by surgical exploration. The diagnostic performance of 4DCT, US, and SeS was compared in terms of sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and accuracy. Additionally, enhancement patterns, false-negative and false-positive 4DCT results, and adenoma location in cases with negative US and SeS findings were analyzed.

Results: One hundred forty-six 4DCT studies were included. Overall, adding 4DCT to US and/or SeS demonstrated significantly higher sensitivity (76.4%) and accuracy (75.7%) compared to US (31.8% and 35.7%) and SeS (26.2% and 31.6%) alone. Specificity was highest for SeS (85.7%), followed by 4DCT (78.3%) and US (72.5%). PPV was high for all modalities, with 4DCT achieving 96.1%, while NPV was low (22.4% for 4DCT, 9.8% for US, 10.3% for SeS). False-negative 4DCT results were mostly due to adenoma adherence to the thyroid. 4DCT identified 48 adenomas missed by US and SeS, with 21 located in ectopic sites.

Conclusion: Adding 4DCT to US and/or SeS provides superior sensitivity and accuracy in detecting parathyroid adenomas compared to US and SeS alone. However, negative results should be interpreted cautiously in patients with clinical and biochemical suspicion for primary hyperparathyroidism.

Critical relevance statement: Adding 4DCT to US and SeS offers superior sensitivity and accuracy in detecting parathyroid adenomas, potentially enhancing clinical decision-making-particularly in challenging cases where prior imaging was negative-though negative results should be interpreted with caution.

Key points: Adding 4D-CT to US and sestamibi scintigraphy (SeS) significantly improves the detection of parathyroid adenomas over US and SeS alone. 4D-CT provides the highest agreement with surgical findings, improving surgical planning. 4D-CT effectively identifies adenomas in ectopic locations, often missed by US and SeS. The most common cause of false-negative 4D-CT results is adenoma adherence to the thyroid, misread as thyroid tissue. Negative imaging results across all modalities warrant cautious interpretation.

Objectives: Our objective was to provide further evidence regarding the diagnostic accuracy of ultrasound-guided attenuation parameter (UGAP) for detecting hepatic steatosis in a selected and homogeneous cohort of patients with impaired glucose tolerance using magnetic resonance imaging-proton density fat fraction (MRI-PDFF) as the standard reference method.

Materials and methods: From October 2023 to March 2024, individuals with impaired glucose tolerance and suspected metabolic dysfunction-associated steatotic liver disease (MASLD) who underwent abdominal ultrasound and MRI-PDFF were enrolled in this prospective study. Multivariable linear regression was performed to assess independent factors associated with UGAP. The correlation between UGAP, and MRI-PDFF was evaluated using the Pearson correlation coefficient. The diagnostic performance of the UGAP for hepatic steatosis was assessed using the area under the receiver operating characteristic curve (AUC).

Results: A total of 193 participants (median age, 35 years; IQR, 25-48 years; 98 male) were evaluated, including 90 participants (47%) with MASLD. Body mass index (BMI), waist circumference, and skin-liver capsule distance were independently associated with UGAP. UGAP was strongly and positively correlated with MRI-PDFF (r = 0.890, p < 0.001). The AUCs of UGAP for diagnosing hepatic steatosis ≥ S1 (MRI-PDFF ≥ 6.4%), ≥ S2 (MRI-PDFF ≥ 17.4%), and ≥ S3 (MRI-PDFF ≥ 22.1%) were 0.980 (95% CI: 0.957-1.000), 0.955 (95% CI: 0.912-0.999), and 0.985 (95% CI: 0.970-1.000), respectively.

Conclusions: UGAP could accurately diagnose hepatic steatosis and estimate the hepatic fat fraction.

Critical relevance statement: UGAP has high diagnostic performance for hepatic steatosis. UGAP offers a method of quantifying liver fat content similar to MRI-PDFF, which is anticipated to be beneficial in the long-term monitoring of patients with impaired glucose tolerance.

Key points: Hepatic steatosis is an early factor that causes liver damage in MASLD. UGAP has an accurate diagnostic performance for hepatic steatosis. UGAP was positively correlated with MRI- proton density fat fraction. Some clinical parameters may affect UGAP measurement.

Objectives: To compare field-of-view optimized and constrained undistorted single-shot (FOCUS), multiplexed sensitivity-encoding (MUSE) and FOCUS-MUSE diffusion-weighted images (DWIs) in orbital imaging quality and staging performance for the patients with thyroid-associated ophthalmopathy (TAO).

Materials and methods: 67 TAOs underwent FOCUS, MUSE and FOCUS-MUSE DWIs. Qualitative (artifacts and geometric distortion, overall image quality, sharpness of boundaries) and quantitative parameters (geometric distortion ratio (GDR), signal-to-noise ratio (SNR), apparent diffusion coefficient (ADC) value, normalized ADC (nADC) value) were assessed. Additionally, nADC values of the extraocular muscles (EOMs) and mean nADC values were compared between active and inactive TAOs. Diagnostic performance was also evaluated.

Results: FOCUS-MUSE DWI exhibited significantly fewer artifacts and geometric distortion, superior overall image quality, enhanced sharpness of boundaries, higher SNR and lower GDR than MUSE and FOCUS DWIs (all p < 0.05). FOCUS-MUSE DWI showed significantly lower ADC values than MUSE (all p < 0.05) and FOCUS DWIs (all p < 0.05, except for that of superior EOM). The nADC values showed no significance among the three DWIs (all p > 0.05), except for that of the superior EOM. Furthermore, active TAOs showed higher nADC values than inactive TAOs in three DWIs (all p < 0.05). The mean nADC value of FOCUS-MUSE DWI (AUC, 0.890; sensitivity, 84.8%; specificity, 77.3%) performed better than that of MUSE (AUC, 0.713; sensitivity, 54.3%; specificity, 80.7%; p < 0.001) and FOCUS DWIs (AUC, 0.730; sensitivity, 47.8%; specificity, 90.9%; p < 0.001) in diagnosing active TAOs.

Conclusions: FOCUS-MUSE DWI provides superior image quality and staging performance in assessing TAO than MUSE and FOCUS DWIs. We recommend its use for evaluating TAO patients in clinical practice.

Critical relevance statement: Field-of-view optimized and constrained undistorted single-shot multiplexed sensitivity-encoding DWI shows superior image quality and staging performance for thyroid-associated ophthalmopathy than other echo-planar imaging-based modified sequences.

Key points: The superiority among different echo-planar imaging-based modified DWIs in thyroid-associated ophthalmopathy remains unclear. Field-of-view optimized and constrained undistorted single-shot multiplexed sensitivity-encoding (FOCUS-MUSE) DWI outperforms MUSE and FOCUS DWIs in imaging quality. Normalized apparent diffusion coefficient values derived from FOCUS-MUSE DWI improve staging performance of thyroid-associated ophthalmopathy.

Objectives: To analyze revision rates after transjugular intrahepatic portosystemic shunt (TIPS) using expanded polytetrafluoroethylene-covered stentgrafts and to identify predictors of shunt revisions.

Materials and methods: This single-center retrospective study included 514 consecutive patients (mean age 56.9 ± 12.7 years; 194 females) with TIPS placement between 2003 and 2021. Follow-up included clinical assessment, laboratory testing, ultrasound, and computed tomography. Reinterventions were categorized by type and technique. Univariable and multivariable Cox regression analyses were performed to identify predictors of shunt dilation and reduction.

Results: A total of 149 patients (28.9%) required TIPS revision: 95 (18.5%) shunt dilation, 42 (8.2%) shunt reduction, and 12 (2.3%) others. Median time to first revision was 2.8 months (3.2 months for dilation, 1.9 months for reduction). Indications for first shunt dilation were persistent or recurrent refractory ascites (n = 61), recurrent variceal bleeding (n = 7), and asymptomatic stenosis or occlusion of the TIPS tract (n = 27). Indications for shunt reduction were hepatic encephalopathy refractory to conservative measures (n = 39) and acute liver failure following TIPS (n = 3). Forty-seven patients (9.1%) underwent two or more reinterventions. Multivariable Cox analysis identified immediate post-TIPS portosystemic pressure gradients > 8 mmHg, prior hepatic encephalopathy, and hepatorenal syndrome prior to TIPS as predictors of mandatory shunt dilation. In contrast, age ≥ 65 years, female gender, serum sodium levels, and a pre-TIPS hepatic hydrothorax were predictive of shunt reduction during revision.

Conclusion: Around one in three patients requires shunt revision. Predictive factors for revision varied by intervention type: shunt dilation was linked to disease severity and portal pressure, whereas reduction was more closely related to the patient's age and gender.

Critical relevance statement: Patients who undergo TIPS require structured, long-term follow-up to identify clinical situations that may necessitate shunt adaptation or other secondary interventions.

Key points: Shunt revision after TIPS occurs in one-third of patients, with prognostic significance. Several independent prognostic factors for both shunt dilation and reduction were identified. Structured long-term follow-up is crucial to identify patients needing shunt revision.

Objective: This study aimed to investigate the reproducibility of ultrasound-derived fat fraction (UDFF) among operators with different experience levels. Furthermore, it seeks to validate the performance of UDFF in detecting hepatic steatosis.

Materials and methods: The study was conducted at three hospitals and involved patients suspected of having metabolic dysfunction-associated steatotic liver disease (MASLD). Two radiologists took UDFF measurements from each participant to find the best place to measure and evaluate reproducibility. Subsequently, the performance of the UDFF measurements from the left and right liver lobes in detecting hepatic steatosis was compared with magnetic resonance proton density fat fraction (MRI-PDFF) results.

Results: A total of 163 patients were examined for UDFF in both left and right liver lobes. The measurement failure rates were 28.8% for the left lobe and 6.5% for the right lobe. The inter-observer reproducibility was high in the right lobe, with intraclass correlation coefficient (ICCs) of 0.89-0.96. Additionally, 80 participants underwent magnetic MRI-PDFF and UDFF examinations, revealing an area under the curve (AUC) of 0.97 for the right lobe and 0.84 for the left lobe.

Conclusion: UDFF measurements in the left lobe exhibited a higher failure rate and less consistency, while measurements in the right lobe had a higher success rate and excellent reproducibility. For suspected MASLD patients, at least two measurements were recommended. Additionally, UDFF from the right lobe is more reliable for detecting fatty liver. However, the study had a small sample size, and future research should include larger, multi-center studies.

Critical relevance statement: Ultrasound-derived fat fraction (UDFF) represents an objective technique characterized by high stability and reproducibility that accurately assesses liver fat content, which is suitable for the screening of metabolic dysfunction-associated steatotic liver disease (MASLD).

Key points: Ultrasound-derived fat fraction (UDFF) obtained from the right lobe demonstrates strong repeatability and reproducibility. UDFF has good diagnostic performance in grading steatosis. UDFF is suitable for the screening of metabolic dysfunction-associated steatotic liver disease (MASLD).