Background and Purpose: Manual segmentation of gynecologic organs and cervical tumors for radiotherapy planning is time-consuming and variable. Automated segmentation on routine T2-weighted magnetic resonance imaging (MRI) remains limited. The aim of this study was to evaluate a lightweight deep learning model for automated segmentation of gynecologic organs and cervical tumors on T2-weighted MRI.Materials and Methods: This work applied a two-stage lightweight deep learning model (PocketNet) to segment the cervix, vagina, uterus, and tumor(s) on T2-weighted MRI in 102 patients with cervical cancer undergoing definitive radiotherapy. Model performance was assessed using the Dice-Sorensen coefficient (DSC) and 95th percentile Hausdorff distance (Haus95) on internal data and validated on an external dataset. A full nnU-Net model trained on the same internal dataset served as a benchmark for segmentation accuracy and computational efficiency.Results: On the institutional dataset, PocketNet achieved mean DSC values exceeding 70% for tumor segmentation and 80% for organ segmentation. External validation on The Cancer Imaging Archive (TCIA) Cervical Cancer Tumor Heterogeneity (CCTH) collection demonstrated the model's robustness, achieving DSC scores of 67.3% for tumor segmentation and 80.8% for organ segmentation. Compared to the PocketNet architecture, nnUNet achieved similar accuracy but required approximately twice the training time, more than 35 times as many parameters, and 40 times more memory for model storage.Conclusion: The PocketNet architecture provides reliable automated segmentation of gynecologic organs and cervical tumors on T2-weighted MRI, with performance comparable to a full-sized nnUNet while requiring substantially less memory and training time, supporting its potential integration into time-sensitive radiotherapy workflows.

Background Accurate assessment of bone mineral density (BMD) is crucial for evaluating bone loss in elderly and oncologic patients. Quantitative computed tomography (QCT) enables noninvasive quantification of patient BMD and can be performed opportunistically during routine CT examinations using either conventional or multi-energy CT (MECT). While both methods are used clinically, MECT can address limitations of conventional CT-based QCT. Purpose This study addresses a critical gap in the literature by systematically evaluating commercially available calibration phantoms for use with MECT, assessing the impact of protocol variations, establishing a calcium-based material decomposition workflow for BMD quantification, validating phantom-derived metrics against patient data, and providing recommendations for integrating quality control (QC) into routine clinical workflows. Methods Five CT QC phantoms containing bone-approximating materials, including calcium (Phantom A), dipotassium phosphate (Phantom B), hydroxyapatite (Phantom C), and calcium carbonate (Phantoms D and E), were evaluated by determining the dual-energy ratio (DER) of inserts. Phantoms were scanned on two identical dual-source CTs at 90/150Sn and 100/150Sn kVp with dose levels of 10 and 20 mGy. Images were reconstructed using filtered back-projection and iterative reconstruction. Phantom DERs were compared with retrospective vertebral measurements from a 10-patient cohort. Results Phantom A provided the most suitable representation of patient data, exhibiting a DER of 1.55 [95%CI:1.54-1.57] at 100/150Sn kVp, compared with 1.44 [95%CI:1.37-1.50] in patients. kVp significantly influenced response, whereas radiation dose and reconstruction approach had minimal effect. Conclusions Phantom selection is critical for MECT-based QCT, and calcium-based phantoms are well-suited for clinical QC and BMD quantification workflows.

To evaluate quantitative imaging biomarkers, including MRI tumor size, apparent diffusion coefficient (ADC), arterial-phase enhancement, and PET maximum standardized uptake value (SUVmax), as prognostic indicators of overall survival (OS), recurrence-free survival (RFS), and treatment response in cervical squamous cell carcinoma (SCC). Fifty patients with biopsy-proven SCC who underwent pre- and post-treatment MRI and FDG-PET/CT were retrospectively analyzed. Tumor dimensions (axial, sagittal), ADC, SUVmax, and arterial enhancement were assessed. Survival was estimated by Kaplan–Meier, and associations with OS and RFS were tested using Cox regression. Logistic regression was used to determine predictors of treatment response according to RECIST and PERCIST criteria. Tumor size, SUVmax, and ADC changed significantly post-treatment (p < 0.001). Persisting arterial enhancement was seen in 35

To compare the diagnostic performance of PET/MRI versus PET/CT for lymph node and metastatic disease detection in patients with gynecologic cancers, and to assess whether PET/MRI offers improved staging accuracy and operational efficiency. 49 patients with known gynecologic malignancies underwent both PET/CT and PET/MRI as part of a prospective study. Imaging protocols followed standard guidelines for each modality, with PET/CT and PET/MRI performed consecutively. Images were independently evaluated by experienced radiologists who assessed primary tumor, nodal, and metastatic disease. Subgroup analysis was performed for cervical cancer patients. Each patient’s time in the imaging department for each scan was recorded, and statistical analysis included calculation of sensitivity, specificity, McNemar testing, and Cohen’s Kappa for inter-rater agreement. This study prospectively confirms prior findings of PET/MRI’s superior nodal detection in gynecologic cancers while providing novel evidence regarding operational efficiency and workflow optimization. PET/MRI demonstrated superior detection of nodal metastases compared to PET/CT, identifying additional adenopathy in 5 patients (p = 0.031) and leading to staging changes in 18 patients within the cohort. In the cervical cancer subset, PET/MRI resulted in more upstaging and downstaging events, though subgroup statistical significance was not reached. PET/MRI also identified more primary tumors missed by PET/CT. The average total imaging time was significantly reduced with PET/MRI (180.3 min) compared to the combined PET/CT and MRI workflow (291.2 min), reflecting a 38.1

OBJECTIVES:To evaluate combined digital breast tomosynthesis and contrast-enhanced mammography (DBT/CEM) for predicting pectoralis muscle invasion. METHODS:This retrospective multi-reader cohort study included research patients who underwent combined DBT/CEM for breast cancer staging and had prepectoral masses. Images were independently reviewed by six fellowship-trained breast radiologists. Diagnostic performance, reader confidence, and inter-reader agreement were calculated for each image type/modality. RESULTS:Among 10 patients with prepectoral masses on DBT/CEM, muscle invasion was present in 3 and absent in 7. The overall diagnostic accuracy of DBT/CEM for PMI was 0.6 (range 0.4-0.9); for predefined radiologic signs it was 0.5-0.7 for low energy (LE) CEM, 0.4-0.7 for DBT, and 0.4-0.8 for recombined (RC) CEM. Muscle deformity on MLO views had the highest accuracy (0.7-0.8). On a scale of 1-3, mean radiologist confidence for combined DBT/CEM was 1.9 (1.5-2.3; SD=0.65). Median confidence ranged from 1.9 for RC to 2.2 for DBT. Per-case reader agreement was poor (K=-0.01) for DBT/CEM; poor to slight (K= -0.13-0.40, median 0.28) for RC; slight to fair (K = 0.04-0.43, median 0.27 and K = 0.02-0.42, median 0.19, respectively) for DBT and LE. In two patients with subpectoral breast implants CEM was accurate in PMI detection, while MRI had one false-positive result. CONCLUSION:Combined DBT/CEM accuracy and inter-reader agreement are suboptimal for PMI evaluation, except in patients with breast implants. RC images marginally improve accuracy compared to LE images but have lowest radiologist confidence. DBT has lowest accuracy but highest confidence. Muscle deformity on MLO view was the most accurate sign. CRITICAL RELEVANCE STATEMENT:Combined DBT/CEM demonstrated suboptimal diagnostic accuracy, reader confidence, and inter-reader agreement for detecting pectoralis muscle invasion (PMI) in prepectoral breast cancer (BC) except for patients with subpectoral breast implants, where recombined images on implant-displaced CEM views performed better than MRI.

Triple-negative breast cancer (TNBC) is a heterogeneous disease with variable response to neoadjuvant systemic therapy (NAST). Patients with pathologic complete response (pCR) following NAST have improved survival. Our goal was to establish readily accessible imaging biomarker to identify which TNBC patients will have pCR. Building on prior favorable results in the literature, we hypothesized that manually measured tumor volume changes at DCE-MRI may predict pCR early during NAST. This prospective study included 287 stage I-III TNBC patients who underwent DCE-MRI at baseline, after two and four cycles of NAST, with pCR accessed at surgery (NCT02276443). Tumor volume and percentage tumor volume reduction were calculated, and their correlation with pCR was evaluated. Our study showed that manually extracted tumor volume changes from DCE-MRI early during NAST were able to predict pCR with high accuracy and can serve as a clinically relevant imaging biomarker for prediction of NAST response in TNBC patients.

Objective: This study aimed to compare the diagnostic accuracy of whole-body PET/MR imaging and contrast-enhanced CT for detecting metastatic disease in patients undergoing surgical resection, using pathology as the reference standard. Materials and Methods: Nineteen patients with suspected metastatic involvement (including four who received neoadjuvant therapy before surgery) underwent both FDG PET/MR and contrast-enhanced CT scans. Imaging was reviewed for metastases at defined sites (e.g., perihepatic region, hepatic parenchyma, mesentery, bowel serosa, colon surface, and nodal basins). Findings on each modality were compared to surgical pathology results per site. Sensitivity, specificity, accuracy, positive predictive value (PPV), and negative predictive value (NPV) were calculated for PET/MR and CT, with pathology as the reference standard. Results: Overall, PET/MR achieved approximately 55.5% sensitivity, 89.5% specificity, 82.5% accuracy, 57.6% positive predictive value (PPV), and 88.6% negative predictive value (NPV). In contrast, CT demonstrated 75.0% sensitivity, 72.3% specificity, 72.9% accuracy, 42% PPV, and 91.5% NPV. No significant correlations were observed between semi-quantitative PET/MR measures, such as SUV or MR ADC values, and patient survival outcomes; therefore, these metrics were excluded from further analysis. Notably, PET/MR imaging findings changed clinical management in 3/6 chemotherapy patients. PET/MR demonstrated greater sensitivity in detecting nodal metastases, 75% compared to CT (25%), and identified small bowel serosal lesions in 1 of 1 case (100% sensitivity) versus none with CT. CT showed slightly higher specificity (81%) for colon serosal involvement than PET/MR (75%). Conclusions: CT demonstrates higher sensitivity, whereas PET/MR offers greater specificity and negative predictive value. When used together, the two modalities may provide a more reliable and comprehensive assessment of metastatic disease.

PURPOSE:To compare image quality and clinical utility of a T2-weighted (T2W) 3-dimensional (3D) fast spin echo (FSE) sequence using deep learning reconstruction (DLR) versus conventional reconstruction for rectal magnetic resonance imaging (MRI). METHODS:The study included 50 patients with rectal cancer who underwent rectal MRI consecutively between July 7, 2020 and January 20, 2021 using a T2W 3D FSE sequence with DLR and conventional reconstruction. Three radiologists reviewed the two sets of images, scoring overall SNR, motion artifacts, and overall image quality on a 3-point scale and indicating clinical preference for DLR or conventional reconstruction based on those three criteria as well as image characterization of bowel wall layer definition, tumor invasion of muscularis propria, residual disease, fibrosis, nodal margin, and extramural venous invasion. RESULTS:Image quality was rated as moderate or good for both DLR and conventional reconstruction for most cases. DLR was preferred over conventional reconstruction in all of the categories except for bowel wall layer definition. CONCLUSION:Both conventional reconstruction and DLR provide acceptable image quality for T2W 3D FSE imaging of rectal cancer. DLR was clinically preferred over conventional reconstruction in almost all categories.

Aim: This study analyzed the associations between various clinical and imaging parameters with overall survival (OS) and recurrence-free survival (RFS) in vulvar cancer. Materials and Methods: A total of 45 patients diagnosed with vulva tumors were retrospectively analyzed. Data were extracted from medical records, including age, tumor size, ADC, SUVmax, and metastases identified through MRI and PET. Survival outcomes were estimated using Kaplan-Meier methods, while associations between variables and survival were assessed using Cox regression. Optimal cut-points for continuous variables were determined using maximally selected rank statistics. Results: The median OS was 9.97 years, with age, tumor size, and SUVmax measurements significantly influencing OS. Optimal cut-points at 4-year survival were established for age 65.9 years, the largest axial dimension of 5.50 cm, craniocaudal dimension of 4.7 cm, SUVmax of 22.0, and ADC value of 1.026 x10-3 mm2/s. Patients with measurements above these cut points typically had worse survival outcomes. Conclusion: Age, Size, and SUVmax predict survival in patients with vulvar cancer.

Suspicious non-calcified mammographic findings have not been evaluated with modern mammographic technique, and the purpose of this work is to compare the likelihood of malignancy for those findings. To do this, 5018 consecutive mammographically guided biopsies performed during 2016–2019 at a large metropolitan, community-based hospital system were retrospectively reviewed. In total, 4396 were excluded for targeting calcifications, insufficient follow-up, or missing data. Thirty-seven of 126 masses (29.4%) were malignant, 44 of 194 asymmetries (22.7%) were malignant, and 77 of 302 architectural distortions (AD, 25.5%) were malignant. The combined likelihood of malignancy was 25.4%. Older age was associated with a higher likelihood of malignancy for each imaging finding type (all p ≤ 0.006), and a possible ultrasound correlation was associated with a higher likelihood of malignancy when all findings were considered together (p = 0.012). Two-view asymmetries were more frequently malignant than one-view asymmetries (p = 0.03). There were two false-negative biopsies (98.7% sensitivity and 100% specificity). In conclusion, the 25.4% likelihood of malignancy confirms the recommendation for biopsy of suspicious, ultrasound-occult, mammographic findings. Mammographically guided biopsies were highly sensitive and specific in this study. Older patient age and a possible ultrasound correlation should raise concern given the increased likelihood of malignancy in those scenarios.

Background While there are clear guidelines regarding chest wall ultrasound in the symptomatic patient, there is conflicting evidence regarding the use of ultrasound in the screening of women post-mastectomy. Objective To assess the utility of screening chest wall ultrasound after mastectomy and to assess features of detected malignancies. Methods This IRB approved, retrospective study evaluates screening US examinations of the chest wall after mastectomy. Asymptomatic women presenting for screening chest wall ultrasound from January 2016 through May 2017 were included. Cases of known active malignancy were excluded. All patients had at least one year of clinical or imaging follow-up. 43 exams (8.5 %) were performed with a history of contralateral malignancy, 465 exams (91.3 %) were performed with a history of ipsilateral malignancy, and one exam (0.2 %) was performed in a patient with bilateral prophylactic mastectomy. Results During the 17-month period, there were 509 screening US in 389 mastectomy patients. 504 (99.0 %) exams were negative/benign. Five exams (1.0 %) were considered suspicious, with recommendation for biopsy, which was performed. Out of 509 exams, 3 (0.6 %) yielded benign results, while 2 (0.39 %) revealed recurrent malignancy, with a 95 % confidence interval (exact binomial) of 0.05 % to 1.41 % for screening ultrasound. Both patients who recurred had previously recurred, and both had initial cancer of lobular histology. Conclusion Of 509 chest wall screening US exams performed in mastectomy, 2 malignancies were detected, and each patient had history of invasive lobular carcinoma and at least one prior recurrence prior to this study, suggesting benefit of screening ultrasound in these populations.

Cervical cancer remains the fourth most common malignancy amongst women worldwide.1 Concurrent chemoradiotherapy (CRT) serves as the mainstay definitive treatment regimen for locally advanced cervical cancers and includes external beam radiation followed by brachytherapy.2 Integral to radiotherapy treatment planning is the routine contouring of both the target tumor at the level of the cervix, associated gynecologic anatomy and the adjacent organs at risk (OARs). However, manual contouring of these structures is both time and labor intensive and associated with known interobserver variability that can impact treatment outcomes. While multiple tools have been developed to automatically segment OARs and the high-risk clinical tumor volume (HR-CTV) using computed tomography (CT) images,3,4,5,6 the development of deep learning-based tumor segmentation tools using routine T2-weighted (T2w) magnetic resonance imaging (MRI) addresses an unmet clinical need to improve the routine contouring of both anatomical structures and cervical cancers, thereby increasing quality and consistency of radiotherapy planning. This work applied a novel deep-learning model (PocketNet) to segment the cervix, vagina, uterus, and tumor(s) on T2w MRI. The performance of the PocketNet architecture was evaluated, when trained on data via five-fold cross validation. PocketNet achieved a mean Dice-Sorensen similarity coefficient (DSC) exceeding 70 tumor segmentation and 80 available dataset from The Cancer Imaging Archive (TCIA) demonstrated the models robustness, achieving DSC scores of 67.3 80.8 variations in contrast protocols, providing reliable segmentation of the regions of interest.

The objective of this study was to compare the quantitative radiomics data between malignant mixed Müllerian tumors (MMMTs) and endometrial carcinoma (EC) and identify texture features associated with overall survival (OS). This study included 61 patients (36 with EC and 25 with MMMTs) and analyzed various radiomic features and gray-level co-occurrence matrix (GLCM) features. These variables and patient clinicopathologic characteristics were compared between EC and MMMTs using the Wilcoxon Rank sum and Fisher’s exact test. The area under the curve of the receiving operating characteristics (AUC ROC) was calculated for univariate analysis in predicting EC status. Logistic regression with elastic net regularization was performed for texture feature selection. This study showed that skewness (p = 0.045) and tumor volume (p = 0.007) significantly differed between EC and MMMTs. The range of cluster shade, the angular variance of cluster shade, and the range of the sum of squares variance were significant predictors of EC status (p ≤ 0.05). The regularized Cox regression analysis identified the “256 Angular Variance of Energy” texture feature as significantly associated with OS independently of the EC/MMMT grouping (p = 0.004). The volume and texture features of the tumor region may help distinguish between EC and MMMTs and predict patient outcomes.

The aim of this study was to determine surgical and clinical outcomes of lobular neoplasia (LN) diagnosed by magnetic resonance imaging (MRI) biopsy, including upgrade to malignancy, and to assess for characteristics associated with upgrade. A single-institution retrospective study, between 2013 and 2022, of patients with histopathological findings of LN via MRI-guided biopsy was performed using an institutional database and review of the electronic medical records. Decision for excision or surveillance was made by a multidisciplinary team per institutional practice. Patient demographics and imaging characteristics were summarized using descriptive analyses. Upgrade was defined as upgrade to cancer on surgical pathology for patients treated with excision or the development of cancer at the biopsy site during surveillance. The Wilcoxon rank-sum test and Fisher’s exact test were used to compare features of the upgraded cohort with the remainder of the group. Ninety-four MRI biopsies diagnosing LN were included. Median age was 57 years (range 37–78 years). Forty-six lesions underwent excision while 48 lesions were surveilled. The upgrade rate was 7.4

OBJECTIVE:To determine the optimal imaging modality for women with high-grade neuroendocrine carcinoma of the cervix.METHODS:Women with high-grade neuroendocrine carcinoma of the cervix who had undergone a computed tomography (CT) scan and combined positron emission tomography with computed tomography (PET/CT) scan within 4 weeks of each other were identified from the NeCTuR Cervical Tumor Registry. One radiologist reviewed all CT scans, and another radiologist reviewed all PET/CT scans. The radiologists denoted the presence or absence of disease at multiple sites. Each radiologist was blinded to prior reports, patient outcomes, and the readings of the other radiologist. With findings on PET/CT used as the gold standard, sensitivity, specificity, and accuracy were calculated for CT scans.RESULTS:Fifty matched CT and PET/CT scans were performed in 41 patients. For detecting primary disease in the cervix, CT scan had a sensitivity of 85%, a specificity of 46%, and an accuracy of 74%. For detecting disease spread to the liver, CT scan had a sensitivity of 80%, a specificity of 89%, and an accuracy of 86%. For detecting disease spread to the lung, CT had a sensitivity of 89%, a specificity of 68%, and an accuracy of 77%. Of the 14 patients who had scans for primary disease work-up, 4 (29%) had a change in their treatment plan due to the PET/CT scan. Had treatment been prescribed on the basis of the CT scan alone, 2 patients would have been undertreated, and 2 would have been overtreated.CONCLUSION:A CT scan is inferior to a PET/CT scan in assessment of metastatic disease in women with high-grade neuroendocrine carcinoma of the cervix. Almost one-third of patients with newly diagnosed high-grade neuroendocrine cervical cancer would have received incorrect therapy had treatment planning been based solely on a CT scan. We recommend a PET/CT scan for both initial work-up and surveillance in women with high-grade neuroendocrine carcinoma of the cervix.

Objectives Evaluate deep learning (DL) to improve the image quality of the PROPELLER (Periodically Rotated Overlapping Parallel Lines with Enhanced Reconstruction technique) for 3 T magnetic resonance imaging of the female pelvis. Methods Three radiologists prospectively and independently compared non-DL and DL PROPELLER sequences from 20 patients with a history of gynecologic malignancy. Sequences with different noise reduction factors (DL 25%, DL 50%, and DL 75%) were blindly reviewed and scored based on artifacts, noise, relative sharpness, and overall image quality. The generalized estimating equation method was used to assess the effect of methods on the Likert scales. Quantitatively, the contrast-to-noise ratio and signal-to-noise ratio (SNR) of the iliac muscle were calculated, and pairwise comparisons were performed based on a linear mixed model. P values were adjusted using the Dunnett method. Interobserver agreement was assessed using the κ statistic. P value was considered statistically significant at less than 0.05. Results Qualitatively, DL 50 and DL 75 were ranked as the best sequences in 86% of cases. Images generated by the DL method were significantly better than non-DL images (P < 0.0001). Iliacus muscle SNR on DL 50 and DL 75 was significantly better than non-DL images (P < 0.0001). There was no difference in contrast-to-noise ratio between the DL and non-DL techniques in the iliac muscle. There was a high percent agreement (97.1%) in terms of DL sequences' superior image quality (97.1%) and sharpness (100%) relative to non-DL images. Conclusion The utilization of DL reconstruction improves the image quality of PROPELLER sequences with improved SNR quantitatively.

OBJECTIVE:The coronavirus disease of 2019 (COVID-19) pandemic disproportionately affected certain vulnerable communities. The purpose of our study was to determine how COVID-19 affected the socioeconomic demographics of breast imaging patients at a major comprehensive cancer center. METHODS:This retrospective cohort study compared female patients who underwent screening mammograms, diagnostic mammograms, breast ultrasound, or breast MRI during the following time periods: prepandemic (February 1, 2018, through February 29, 2020), acute pandemic (March 1, 2020, through June 30, 2020), subacute pandemic (August 1, 2020, through December 31, 2020), and chronic pandemic (January 1, 2021, through June 30, 2022). Statistics were performed using the generalized estimating equations approach. RESULTS:A total of 74,398 female patients (mean age, 55.6 ± 12.4 years) underwent 238,776 total breast imaging examinations. For screening mammograms, Hispanics represented 27.1% (9,197 of 33,960) of patients in the prepandemic time period compared with 16.7% (604 of 3,621) in the acute pandemic time period, 18.7% (1,835 of 9,830) in the subacute pandemic time period, and 24.3% (7,492 of 30,869) in the chronic pandemic time period (all P < .0001). Self-pay patients saw similar declines for screening mammograms during the same time periods: 21.7% (7,375 of 33,960), 7.9% (286 of 3,621), 9.5% (933 of 9,830), and 17.4% (5,357 of 30,869), respectively (all P < .0001, compared with the prepandemic time period). Similarly dramatic trends were not observed for race or other imaging examinations. DISCUSSION:At our cancer center, Hispanics and self-pay patients were disproportionately affected by the COVID-19 pandemic. Strategies to improve health inequities are needed.

Cholangiocarcinoma (CCA) is an aggressive solid tumour with a 5-year survival rate ranging from 7% to 20%. It is, therefore, urgent to identify novel biomarkers and therapeutic targets to improve the outcomes of patients with CCA. SPRY-domain containing protein 4 (SPRYD4) contains SPRY domains that modulate protein-protein interaction in various biological processes; however, its role in cancer development is insufficiently explored. This study is the first to identify that SPRYD4 is downregulated in CCA tissues using multiple public datasets and a CCA cohort. Furthermore, the low expression of SPRYD4 was significantly associated with unfavourable clinicopathological characteristics and poor prognosis in patients with CCA, indicating that SPRYD4 could be a prognosis indicator of CCA. In vitro experiments revealed that SPRYD4 overexpression inhibited CCA cells proliferation and migration, whereas the proliferative and migratory capacity of CCA cells was enhanced after SPRYD4 deletion. Moreover, flow cytometry showed that SPRYD4 overexpression triggered the S/G2 cell phase arrest and promoted apoptosis in CCA cells. Furthermore, the tumour-inhibitory effect of SPRYD4 was validated in vivo using xenograft mouse models. SPRYD4 also showed a close association with tumour-infiltrating lymphocytes and important immune checkpoints including PD1, PD-L1 and CTLA4 in CCA. In conclusion, this study elucidated the role of SPRYD4 during CCA development and highlighted SPRYD4 as a novel biomarker and tumour suppressor in CCA.