Objective
To develop and evaluate the performance of U-Net for fully automated localization and segmentation of cervical tumors in magnetic resonance (MR) images and the robustness of extracting ...apparent diffusion coefficient (ADC) radiomics features.
Methods
This retrospective study involved analysis of MR images from 169 patients with cervical cancer stage IB–IVA captured; among them, diffusion-weighted (DW) images from 144 patients were used for training, and another 25 patients were recruited for testing. A U-Net convolutional network was developed to perform automated tumor segmentation. The manually delineated tumor region was used as the ground truth for comparison. Segmentation performance was assessed for various combinations of input sources for training. ADC radiomics were extracted and assessed using Pearson correlation. The reproducibility of the training was also assessed.
Results
Combining b0, b1000, and ADC images as a triple-channel input exhibited the highest learning efficacy in the training phase and had the highest accuracy in the testing dataset, with a dice coefficient of 0.82, sensitivity 0.89, and a positive predicted value 0.92. The first-order ADC radiomics parameters were significantly correlated between the manually contoured and fully automated segmentation methods (
p
< 0.05). Reproducibility between the first and second training iterations was high for the first-order radiomics parameters (intraclass correlation coefficient = 0.70–0.99).
Conclusion
U-Net-based deep learning can perform accurate localization and segmentation of cervical cancer in DW MR images. First-order radiomics features extracted from whole tumor volume demonstrate the potential robustness for longitudinal monitoring of tumor responses in broad clinical settings.
Summary
U-Net-based deep learning can perform accurate localization and segmentation of cervical cancer in DW MR images.
Key Points
•
U-Net-based deep learning can perform accurate fully automated localization and segmentation of cervical cancer in diffusion-weighted MR images.
•
Combining b0, b1000, and apparent diffusion coefficient (ADC) images exhibited the highest accuracy in fully automated localization.
• First-order radiomics feature extraction from whole tumor volume was robust and could thus potentially be used for longitudinal monitoring of treatment responses.
Purpose
To compare the diagnostic accuracy of contrast‐enhanced (CE) magnetic resonance imaging (MRI) and diffusion‐weighted MRI (DWI) in the differentiation between uterine leiomyosarcoma (LMS) / ...smooth muscle tumor with uncertain malignant potential (STUMP) and benign leiomyoma.
Materials and Methods
A consecutive cohort of 8 LMS/STUMP and 25 benign leiomyomas underwent pelvic MRI exam at 3T. Two radiologists independently evaluated images based on CE‐MRI (central nonenhancement at equilibrial phase) and DWI (hyperintensity on b = 1000 s/mm2 and hypointensity on apparent diffusion coefficients ADC map). The ADC values were calculated from b = 0 and 1000 s/mm2.
Results
CE‐MRI yielded a significantly superior diagnostic accuracy (0.94 vs. 0.52) and a significantly higher specificity (0.96 vs. 0.36) than DWI (P < 0.05 for both), and remained a comparably high sensitivity as DWI (0.88 vs. 1.00). A combination of DWI and ADC value <1.08 × 10−3 mm2/s (determined by receiver operating characteristic analysis) improved diagnostic accuracy, sensitivity, and specificity of DWI to 0.88, 0.88, and 0.88, respectively, by post‐hoc analysis based on the same study cohort.
Conclusion
For prospective differentiation between uterine LMS/STUMP and benign leiomyoma, CE‐MRI can provide accurate information and is preferable to DWI. Combination of DWI and ADC values can achieve a comparable diagnostic accuracy to CE‐MRI. J. Magn. Reson. Imaging 2016;43:333–342.
Objectives
To evaluate the clinical impact of a deep learning system (DLS) for automated detection of pulmonary nodules on computed tomography (CT) images as a second reader.
Methods
This ...single-centre retrospective study screened 21,150 consecutive body CT studies from September 2018 to February 2019. Pulmonary nodules detected by the DLS on axial CT images but not mentioned in initial radiology reports were flagged. Flagged images were scored by four board-certificated radiologists each with at least 5 years of experience. Nodules with scores of 2 (understandable miss) or 3 (should not be missed) were then categorised as unlikely to be clinically significant (2a or 3a) or likely to be clinically significant (2b or 3b) according to the 2017 Fleischner guidelines for pulmonary nodules. The miss rate was defined as the total number of studies receiving scores of 2 or 3 divided by total screened studies.
Results
Among 172 nodules flagged by the DLS, 60 (35%) missed nodules were confirmed by the radiologists. The nodules were further categorised as 2a, 2b, 3a, and 3b in 24, 14, 10, and 12 studies, respectively, with an overall positive predictive value of 35%. Missed pulmonary nodules were identified in 0.3% of all CT images, and one-third of these lesions were considered clinically significant.
Conclusions
Use of DLS-assisted automated detection as a second reader can identify missed pulmonary nodules, some of which may be clinically significant.
Clinical relevance/application.
Use of DLS to help radiologists detect pulmonary lesions may improve patient care.
Key Points
•
DLS-assisted automated detection as a second reader is feasible in a large consecutive cohort.
•
Performance of combined radiologists and DLS was better than DLS or radiologists alone.
•
Pulmonary nodules were missed more frequently in abdomino-pelvis CT than the thoracic CT.
Background: Predicting suicide is a pressing issue among older adults; however, predicting its risk is difficult. Capitalizing on the recent development of machine learning, considerable progress has ...been made in predicting complex behavior such as suicide. As depression remained the strongest risk for suicide, we aimed to apply deep learning algorithms to identify suicidality in a group with late‐life depression (LLD).
Methods: We enrolled 83 patients with LLD, 35 of which were non‐suicidal and 48 were suicidal, including 26 with only suicidal ideation and 22 with past suicide attempts, for resting‐state functional magnetic resonance imaging (MRI). Cross‐sample entropy (CSE) analysis was conducted to examine the complexity of MRI signals among brain regions. Three‐dimensional (3D) convolutional neural networks (CNNs) were used, and the classification accuracy in each brain region was averaged to predict suicidality after sixfold cross‐validation.
Results: We found brain regions with a mean accuracy above 75% to predict suicidality located mostly in default mode, fronto‐parietal, and cingulo‐opercular resting‐state networks. The models with right amygdala and left caudate provided the most reliable accuracy in all cross‐validation folds, indicating their neurobiological importance in late‐life suicide.
Conclusion: Combining CSE analysis and the 3D CNN, several brain regions were found to be associated with suicidality.
Predicting suicide in older adults is difficulty. Using machine learning, we can predict suicidality from the certain brain regions' complexity of the resting state fMRI data.
Objectives/Hypothesis
Plasma Epstein‐Barr virus (EBV) DNA concentrations predict prognosis in patients with nasopharyngeal carcinoma (NPC). Recent evidence also indicates that intratumor ...heterogeneity on F‐18 fluorodeoxyglucose positron emission tomography (18F‐FDG PET) scans is predictive of treatment outcomes in different solid malignancies. Here, we sought to investigate the prognostic value of heterogeneity parameters in patients with primary NPC.
Study Design
Retrospective cohort study.
Methods
We examined 101 patients with primary NPC who underwent pretreatment 18F‐FDG PET/computed tomography. Circulating levels of EBV DNA were measured in all participants. The following PET heterogeneity parameters were collected: histogram‐based heterogeneity parameters, second‐order texture features (uniformity, contrast, entropy, homogeneity, dissimilarity, inverse difference moment), and higher‐order (coarseness, contrast, busyness, complexity, strength) texture features.
Results
The median follow‐up time was 5.14 years. Total lesion glycolysis (TLG), tumor heterogeneity measured by histogram‐based parameter skewness, and the majority of second‐order or higher‐order texture features were significantly associated with overall survival (OS) and/or recurrence‐free survival (RFS). In multivariate analysis, age (P =.005), EBV DNA load (P = .0002), and uniformity (P = .001) independently predicted OS. Only skewness retained the independent prognostic significance for RFS. Tumor stage, standardized uptake value, or TLG did not show an independent association with survival endpoints. The combination of uniformity, EBV DNA load, and age resulted in a more reliable prognostic stratification (P < .001).
Conclusions
Tumor heterogeneity is superior to traditional PET parameters for predicting outcomes in primary NPC. The combination of uniformity with EBV DNA load can improve prognostic stratification in this clinical entity.
Level of Evidence
4 Laryngoscope, 127:E22–E28, 2017
Purpose
To investigate the biological meaning of apparent diffusion coefficient (ADC) values in tumors following radiotherapy.
Materials and Methods
Five mice bearing TRAMP‐C1 tumor were ...half‐irradiated with a dose of 15 Gy. Diffusion‐weighted images, using multiple b‐values from 0 to 3000 s/mm2, were acquired at 7T on day 6. ADC values calculated by a two‐point estimate and monoexponential fitting of signal decay were compared between the irradiated and nonirradiated regions of the tumor. Pixelwise ADC maps were correlated with histological metrics including nuclear counts, nuclear sizes, nuclear spaces, cytoplasmic spaces, and extracellular spaces.
Results
As compared with the nonirradiated region, the irradiated region exhibited significant increases in ADC, extracellular space, and nuclear size, and a significant decrease in nuclear counts (P < 0.001 for all). Optimal ADC to differentiate the irradiated from nonirradiated regions was achieved at a b‐value of 800 s/mm2 by the two‐point method and monoexponential curve fitting. ADC positively correlated with extracellular spaces (r = 0.74) and nuclear sizes (r = 0.72), and negatively correlated with nuclear counts (r = –0.82, P < 0.001 for all).
Conclusion
As a radiomic biomarker, ADC maps correlating with histological metrics pixelwise could be a means of evaluating tumor heterogeneity and responses to radiotherapy.
Level of Evidence: 1
Technical Efficacy: Stage 2
J. MAGN. RESON. IMAGING 2017;46:483–489
Human papillomavirus (HPV) is an oncogenic virus causing oropharyngeal cancers and resulting in a favorable outcome after the treatment. The role of HPV in oral cavity squamous cell carcinoma (OSCC) ...remains ambiguous.
This study aimed to examine the effect of HPV infection on disease control among patients with OSCC following radical surgery with radiation-based adjuvant therapy.
We prospectively followed 173 patients with advanced OSCC (96% were stage III/IV) who had undergone radical surgery and adjuvant therapy between 2004 and 2006. They were followed between surgery and death or up to 60 months. Surgical specimens were examined using a PCR-based HPV blot test. The primary endpoints were the risk of relapse and the time to relapse; the secondary endpoints were disease-free survival, disease-specific survival, and overall survival.
The prevalence of HPV-positive OSCC was 22%; HPV-16 (9%) and HPV-18 (7%) were the genotypes most commonly encountered. Solitary HPV-16 infection was a poor predictor of 5-year distant metastases (hazard ratio, 3.4; 95% confidence interval, 1.4-8.0; P = 0.005), disease-free survival (P = 0.037), disease-specific survival (P = 0.006), and overall survival (P = 0.010), whereas HPV-18 infection had no impact on 5-year outcomes. The rate of 5-year distant metastases was significantly higher in the HPV-16 or level IV/V metastasis group compared with both the extracapsular spread or tumor depth ≥ 11-mm group and patients without risk factors (P<0.001).
HPV infections in advanced OSCC patients are not uncommon and clinically relevant. Compared with HPV-16-negative advanced OSCC patients, those with a single HPV-16 infection are at higher risk of distant metastases and poor survival despite undergoing radiation-based adjuvant therapy and require a more aggressive adjuvant treatment and a more thorough follow-up.
Background
This study investigated the accuracy in achieving proper lower limb alignment and component positions after total knee replacement (TKR) with image‐free and image‐based robotic‐assisted ...TKR.
Methods
A total of 129 patients (166 knees) suffering from end‐stage knee arthritis who underwent TKA operated by robotic‐assisted surgery between the years 2018 and mid‐2021 were recruited. Radiological outcomes were compared between image‐free and image‐based robotic‐assisted surgical systems.
Results
There were significant differences between the two robotic systems when comparing the mean planned component alignment and the mean measured alignment on radiographs, in which the image‐free robotic‐assisted system was more varus, whereas the image‐based robotic‐assisted system was more valgus for both the mean femoral and tibial component coronal alignment (p < 0.001). For tibial component sagittal alignment, the image‐based group had a larger deviation from the planned posterior slope (p < 0.001).
Conclusion
Image‐free and image‐based robotic assisted TKR had differing accuracy in femoral and tibial alignment.
Objectives
Peripheral arterial disease (PAD) is characterised by arterial occlusion and fibrosis in the lower extremities. Extracellular volume matrix fraction (ECV) is a biomarker of skeletal muscle ...fibrosis, but has not been applied to the lower extremities with PAD. This study investigated the clinical feasibility of using ECV for calf muscle fibrosis quantification by comparing normal controls (NC) and PAD patients.
Methods
From October 2016 to December 2017, we recruited patients with PAD, and patients with head and neck cancer receiving fibular flap as NC group. All participants underwent magnetic resonance imaging (MRI) to determine the ECV of the calves and the differences between the NC and PAD groups. ECV was calculated from T1 values at steady-state equilibrium, defined as the point in time after contrast agent injection when the variance of T1 relaxation time in blood and muscle becomes less than 5%.
Results
A total of 46 patients (18 in the NC group and 28 in the PAD group) were recruited. Steady-state equilibrium was reached at 11–12 min after contrast agent injection. The NC group had significantly lower mean ECV than the PAD group (12.71% vs. 31.92%, respectively,
p
< 0.001). In the PAD group, the mean ECV was slightly lower in patients with collateral vessels than in those without (26.58% vs. 34.88%, respectively,
p
= 0.047).
Conclusion
Evaluation of skeletal fibrosis in PAD using ECV is feasible. ECV can help identify PAD patients with collateral vessel formation and lay the foundation for future research in PAD management.
Key Points
• Steady-state equilibrium for ECV measurement of the lower limbs can be reached at around 11–12 min.
• Quantification of lower limb muscle fibrosis by measuring ECV is clinically feasible and can be used to differentiate between patients with PAD and histologically proven normal controls.
• ECV can differentiate PAD patients with or without visible collateral vessels, further expanding its role in identifying the presence of collateral supply in clinical decision-making.