Diagnostic Accuracy of CT for Metastatic Epidural Spinal Cord Compression

Background: Early diagnosis of metastatic epidural spinal cord compression (MESCC) is vital to expedite therapy and prevent paralysis. Staging CT is performed routinely in cancer patients and presents an opportunity for earlier diagnosis. Methods: This retrospective study included 123 CT scans from 101 patients who underwent spine MRI within 30 days, excluding 549 CT scans from 216 patients due to CT performed post-MRI, non-contrast CT, or a gap greater than 30 days between modalities. Reference standard MESCC gradings on CT were provided in consensus via two spine radiologists (11 and 7 years of experience) analyzing the MRI scans. CT scans were labeled using the original reports and by three radiologists (3, 13, and 14 years of experience) using dedicated CT windowing. Results: For normal/none versus low/high-grade MESCC per CT scan, all radiologists demonstrated almost perfect agreement with kappa values ranging from 0.866 (95% CI 0.787–0.945) to 0.947 (95% CI 0.899–0.995), compared to slight agreement for the reports (kappa = 0.095, 95%CI −0.098–0.287). Radiologists also showed high sensitivities ranging from 91.51 (95% CI 84.49–96.04) to 98.11 (95% CI 93.35–99.77), compared to 44.34 (95% CI 34.69–54.31) for the reports. Conclusion: Dedicated radiologist review for MESCC on CT showed high interobserver agreement and sensitivity compared to the current standard of care.

Improved Productivity Using Deep Learning-assisted Reporting for Lumbar Spine MRI

Background Lumbar spine MRI studies are widely used for back pain assessment. Interpretation involves grading lumbar spinal stenosis, which is repetitive and time consuming. Deep learning (DL) could provide faster and more consistent interpretation. Purpose To assess the speed and interobserver agreement of radiologists for reporting lumbar spinal stenosis with and without DL assistance. Materials and Methods In this retrospective study, a DL model designed to assist radiologists in the interpretation of spinal canal, lateral recess, and neural foraminal stenoses on lumbar spine MRI scans was used. Randomly selected lumbar spine MRI studies obtained in patients with back pain who were 18 years and older over a 3-year period, from September 2015 to September 2018, were included in an internal test data set. Studies with instrumentation and scoliosis were excluded. Eight radiologists, each with 2-13 years of experience in spine MRI interpretation, reviewed studies with and without DL model assistance with a 1-month washout period. Time to diagnosis (in seconds) and interobserver agreement (using Gwet κ) were assessed for stenosis grading for each radiologist with and without the DL model and compared with test data set labels provided by an external musculoskeletal radiologist (with 32 years of experience) as the reference standard. Results Overall, 444 images in 25 patients (mean age, 51 years ± 20 [SD]; 14 women) were evaluated in a test data set. DL-assisted radiologists had a reduced interpretation time per spine MRI study, from a mean of 124-274 seconds (SD, 25-88 seconds) to 47-71 seconds (SD, 24-29 seconds) (P < .001). DL-assisted radiologists had either superior or equivalent interobserver agreement for all stenosis gradings compared with unassisted radiologists. DL-assisted general and in-training radiologists improved their interobserver agreement for four-class neural foraminal stenosis, with κ values of 0.71 and 0.70 (with DL) versus 0.39 and 0.39 (without DL), respectively (both P < .001). Conclusion Radiologists who were assisted by deep learning for interpretation of lumbar spinal stenosis on MRI scans showed a marked reduction in reporting time and superior or equivalent interobserver agreement for all stenosis gradings compared with radiologists who were unassisted by deep learning. © RSNA, 2022 Online supplemental material is available for this article. See also the editorial by Hayashi in this issue.

Infiltrative pattern of metastatic invasive lobular breast carcinoma in the abdomen: a pictorial review

Invasive lobular carcinoma (ILC) has a greater tendency to metastasize to the peritoneum, retroperitoneum, and gastrointestinal (GI) tract as compared to invasive carcinoma of no special type (NST). Like primary ILC in the breast, ILC metastases are frequently infiltrative and hypometabolic, rather than mass forming and hypermetabolic in nature. This renders them difficult to detect on conventional and metabolic imaging studies. As a result, intra-abdominal ILC metastases are often detected late, with patients presenting with clinical complications such as liver failure, hydronephrosis, or bowel obstruction. In patients with known history of ILC, certain imaging features are very suggestive of infiltrative metastatic ILC. These include retroperitoneal or peritoneal nodularity and linitis plastica appearance of the bowel. Recognition of linitis plastica on imaging should prompt deep or repeat biopsies. In this pictorial review, the authors aim to familiarize readers with imaging features and pitfalls for evaluation of intra-abdominal metastatic ILC. Awareness of these will allow the radiologist to assess these patients with a high index of suspicion and aid detection of metastatic disease. Also, this can direct histopathology and immunohistochemical staining to obtain the correct diagnosis in suspected metastatic disease.

Detection of Pneumothorax with Deep Learning Models: Learning From Radiologist Labels vs Natural Language Processing Model Generated Labels

RATIONALE AND OBJECTIVES

To compare the performance of pneumothorax deep learning detection models trained with radiologist versus natural language processing (NLP) labels on the NIH ChestX-ray14 dataset.

MATERIALS AND METHODS

The ChestX-ray14 dataset consisted of 112,120 frontal chest radiographs with 5302 positive and 106, 818 negative labels for pneumothorax using NLP (dataset A). All 112,120 radiographs were also inspected by 4 radiologists leaving a visually confirmed set of 5,138 positive and 104,751 negative for pneumothorax (dataset B). Datasets A and B were used independently to train 3 convolutional neural network (CNN) architectures (ResNet-50, DenseNet-121 and EfficientNetB3). All models' area under the receiver operating characteristic curve (AUC) were evaluated with the official NIH test set and an external test set of 525 chest radiographs from our emergency department.

RESULTS

There were significantly higher AUCs on the NIH internal test set for CNN models trained with radiologist vs NLP labels across all architectures. AUCs for the NLP/radiologist-label models were 0.838 (95%CI:0.830, 0.846)/0.881 (95%CI:0.873,0.887) for ResNet-50 (p = 0.034), 0.839 (95%CI:0.831,0.847)/0.880 (95%CI:0.873,0.887) for DenseNet-121, and 0.869 (95%CI: 0.863,0.876)/0.943 (95%CI: 0.939,0.946) for EfficientNetB3 (p ≤0.001). Evaluation with the external test set also showed higher AUCs (p <0.001) for the CNN models trained with radiologist versus NLP labels across all architectures. The AUCs for the NLP/radiologist-label models were 0.686 (95%CI:0.632,0.740)/0.806 (95%CI:0.758,0.854) for ResNet-50, 0.736 (95%CI:0.686, 0.787)/0.871 (95%CI:0.830,0.912) for DenseNet-121, and 0.822 (95%CI: 0.775,0.868)/0.915 (95%CI: 0.882,0.948) for EfficientNetB3.

CONCLUSION

We demonstrated improved performance and generalizability of pneumothorax detection deep learning models trained with radiologist labels compared to models trained with NLP labels.

Deep Learning Systems for Pneumothorax Detection on Chest Radiographs: A Multicenter External Validation Study

PURPOSE

To assess the generalizability of a deep learning pneumothorax detection model on datasets from multiple external institutions and examine patient and acquisition factors that might influence performance.

MATERIALS AND METHODS

In this retrospective study, a deep learning model was trained for pneumothorax detection by merging two large open-source chest radiograph datasets: ChestX-ray14 and CheXpert. It was then tested on six external datasets from multiple independent institutions (labeled A-F) in a retrospective case-control design (data acquired between 2016 and 2019 from institutions A-E; institution F consisted of data from the MIMIC-CXR dataset). Performance on each dataset was evaluated by using area under the receiver operating characteristic curve (AUC) analysis, sensitivity, specificity, and positive and negative predictive values, with two radiologists in consensus being used as the reference standard. Patient and acquisition factors that influenced performance were analyzed.

RESULTS

The AUCs for pneumothorax detection for external institutions A-F were 0.91 (95% CI: 0.88, 0.94), 0.97 (95% CI: 0.94, 0.99), 0.91 (95% CI: 0.85, 0.97), 0.98 (95% CI: 0.96, 1.0), 0.97 (95% CI: 0.95, 0.99), and 0.92 (95% CI: 0.90, 0.95), respectively, compared with the internal test AUC of 0.93 (95% CI: 0.92, 0.93). The model had lower performance for small compared with large pneumothoraces (AUC, 0.88 [95% CI: 0.85, 0.91] vs AUC, 0.96 [95% CI: 0.95, 0.97]; P = .005). Model performance was not different when a chest tube was present or absent on the radiographs (AUC, 0.95 [95% CI: 0.92, 0.97] vs AUC, 0.94 [95% CI: 0.92, 0.05]; P > .99).

CONCLUSION

A deep learning model trained with a large volume of data on the task of pneumothorax detection was able to generalize well to multiple external datasets with patient demographics and technical parameters independent of the training data.Keywords: Thorax, Computer Applications-Detection/DiagnosisSee also commentary by Jacobson and Krupinski in this issue.Supplemental material is available for this article.©RSNA, 2021.

Correct determination of the enhancement curve is critical to ensure accurate diagnosis using the Kaiser score as a clinical decision rule for breast MRI

OBJECTIVES

the Kaiser score is increasingly recognized as a valuable tool to improve breast MRI interpretation. Contrast enhancement kinetics are the second most important diagnostic criterion, thus defining the curve type plays a crucial role in Kaiser score assessment. We investigate whether the timepoint used to determine the initial enhancement (earlyor peak) for the signal-intensity time curve analysis affects the diagnostic performance of the Kaiser score.

METHODS

This IRB-approved, retrospective, single-center study included 70 consecutives histologically verified breast MRI cases. Two off-site breast radiologists independently read all examinations using the Kaiser score, assessing the initial enhancement using three approaches: -first (1 st), second (2nd) and peak (maximum) of either 1 st or 2nd post-contrast timepoints. The initial enhancement was then compared to the last timepoint (delayed enhancement) to determine the curve type. Visual assessment of curve types was used for this study. Diagnostic performance was evaluated by receiver operating characteristics (ROC) analysis.

RESULTS

Kaiser score reading results using the peak enhancement of either the first or second timepoint performed significantly better than the other approaches (P < 0.05, respectively) and specifically achieved higher sensitivity. Diagnostic accuracy (AUC area under the curve) ranged between 85.4 % and 91.6 %, without significant differences between the two readers (P < 0.5).

CONCLUSIONS

Diagnostic performance of the Kaiser score is significantly influenced by how the initial enhancement timepoint is determined. Peak enhancement should be used as initial timepoint to avoid pitfalls due to timing or physiological differences.

Ultrasound Guided Optoacoustic Tomography in Assessment of Tumor Margins for Lumpectomies

PURPOSE: To determine the accuracy of a handheld ultrasound-guided optoacoustic tomography (US-OT) probe developed for human deep-tissue imaging in ex vivo assessment of tumor margins postlumpectomy. METHODS: A custom-built two-dimensional (2D) US-OT–handheld probe was used to scan 15 lumpectomy breast specimens. Optoacoustic signals acquired at multiple wavelengths between 700 and 1100 nm were reconstructed using model linear algorithm, followed by spectral unmixing for lipid and deoxyhemoglobin (Hb). Distribution maps of lipid and Hb on the anterior, posterior, superior, inferior, medial, and lateral margins of the specimens were inspected for margin involvement, and results were correlated with histopathologic findings. The agreement in tumor margin assessment between US-OT and histopathology was determined using the Bland–Altman plot. Accuracy, sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) of margin assessment using US-OT were calculated. RESULTS: Ninety margins (6 × 15 specimens) were assessed. The US-OT probe resolved blood vessels and lipid up to a depth of 6 mm. Negative and positive margins were discriminated by marked differences in the distribution patterns of lipid and Hb. US-OT assessments were concordant with histopathologic findings in 87 of 89 margins assessed (one margin was uninterpretable and excluded), with diagnostic accuracy of 97.9% (kappa = 0.79). The sensitivity, specificity, PPV, and NPV were 100% (4/4), 97.6% (83/85), 66.7% (4/6), and 100% (83/83), respectively. CONCLUSION: US-OT was capable of providing distribution maps of lipid and Hb in lumpectomy specimens that predicted tumor margins with high sensitivity and specificity, making it a potential tool for intraoperative tumor margin assessment.

Convolutional Neural Networks for Automated Fracture Detection and Localization on Wrist Radiographs

PURPOSE

To demonstrate the feasibility and performance of an object detection convolutional neural network (CNN) for fracture detection and localization on wrist radiographs.

MATERIALS AND METHODS

Institutional review board approval was obtained with waiver of consent for this retrospective study. A total of 7356 wrist radiographic studies were extracted from a hospital picture archiving and communication system. Radiologists annotated all radius and ulna fractures with bounding boxes. The dataset was split into training (90%) and validation (10%) sets and used to train fracture localization models for frontal and lateral images. Inception-ResNet Faster R-CNN architecture was implemented as a deep learning model. The models were tested on an unseen test set of 524 consecutive emergency department wrist radiographic studies with two radiologists in consensus as the reference standard. Per-fracture, per-image (ie, per-view), and per-study sensitivity and specificity were determined. Area under the receiver operating characteristic curve (AUC) analysis was performed.

RESULTS

The model detected and correctly localized 310 (91.2%) of 340 and 236 (96.3%) of 245 of all radius and ulna fractures on the frontal and lateral views, respectively. The per-image sensitivity, specificity, and AUC were 95.7% (95% confidence interval [CI]: 92.4%, 97.8%), 82.5% (95% CI: 77.4%, 86.8%), and 0.918 (95% CI: 0.894, 0.941), respectively, for the frontal view and 96.7% (95% CI: 93.6%, 98.6%), 86.4% (95% CI: 81.9%, 90.2%), and 0.933 (95% CI: 0.912, 0.954), respectively, for the lateral view. The per-study sensitivity, specificity, and AUC were 98.1% (95% CI: 95.6%, 99.4%), 72.9% (95% CI: 67.1%, 78.2%), and 0.895 (95% CI: 0.870, 0.920), respectively.

CONCLUSION

The ability of an object detection CNN to detect and localize radius and ulna fractures on wrist radiographs with high sensitivity and specificity was demonstrated.© RSNA, 2019.

Diagnostic utility of whole body CT scanning in patients with unexplained weight loss

BACKGROUND

Unexplained weight loss is a non-specific complaint with myriad potential etiologies. Increasingly, whole body CT studies are being performed in patients with unexplained weight loss to exclude organic etiologies such as malignancy. Our study aims to assess the diagnostic accuracy and yield of whole body CT in these patients.

METHODS AND MATERIALS

Patients who had a whole body CT scan for investigation of unexplained weight loss as their primary complaint from 2009-2012 were retrospectively reviewed. CT scans were classified into 4 categories: (1) Definite/highly suspicious for underlying organic cause (2) Indeterminate for underlying organic cause (3) No findings accounting for weight loss and only incidental findings and (4) Normal study. Scan findings were correlated with the final diagnosis after all investigations. Univariate logistic regression was performed to determine associations between patient's baseline variables and positive CT scan findings.

RESULTS

Of 301 eligible patients during the study period, 101 patients were excluded due to known history of malignancy, inadequate follow-up or inadequate scan technique. 200 patients were included in the final analyses. The sensitivity, specificity, positive predictive value, negative predictive value and accuracy of CT for organic pathology were 72.0%, 90.7%, 87.0%, 78.9% and 82.0% respectively. Additional symptoms, abnormal physical examinations, anemia, and raised tumor markers were significantly correlated with positive CT findings. Overall, the diagnostic yield of whole body CT scan for patients with unexplained weight loss was 33.5%.

CONCLUSIONS

Whole body CT imaging may be a useful investigation for the noninvasive workup of patients with unexplained weight loss, with diagnostic yield of 33.5% and good sensitivity, specificity, positive and negative predictive values for organic etiologies.

CT features in surgically proven cases of ovarian torsion-a pictorial review

Ovarian torsion is a surgical emergency characterized by a partial or complete rotation of the ovarian vascular pedicle, causing obstruction to venous outflow followed by arterial inflow. Clinically, ovarian torsion frequently mimics other causes of acute pelvic pain such as appendicitis, diverticulitis, renal colic etc. Ultrasonography is the first-line imaging modality of choice for evaluation of ovarian torsion. MRI is useful as a problem-solving tool in patients with equivocal or non-diagnostic ultrasonography studies. CT is ordinarily not utilized in a young female with suspected ovarian torsion due to the radiation dose. However, the significant expansion in use of CT imaging in emergency departments for female patients presenting with acute abdominal pain has increased the likelihood that ovarian torsion may be first seen on CT. In addition, a non-specific clinical presentation may lead to an initial imaging with CT rather than ultrasonography. Ultrasound features of the ovarian torsion are well known and sufficiently described across literature as compared with the CT scan findings. In view of the increasing usage of CT as the modality of choice in emergency settings, it is imperative for the radiologist to familiarize with the CT features of ovarian torsion. An early correct diagnosis by the radiologist in clinically unsuspected cases, facilitating a prompt surgery to restore the ovarian blood flow can prevent permanent irreversible damage. There is limited published data available on the CT features of ovarian torsion. This pictorial essay illustrates CT findings with histological correlation of surgically proven ovarian torsion in our institution. These patients were primarily investigated with CT scan for acute pelvic pain ascribed to non-gynaecological causes such as bowel or urinary tract lesions.

Papillary lesions of the breast: imaging findings and diagnostic challenges

Papillary breast lesions encompass a wide spectrum of pathologies ranging from benign lesions, such as solitary intraductal papilloma, to the uncommon papillary carcinoma. These lesions have various clinical presentations and diverse radiological features. Differentiating benign and malignant papillary lesions based on imaging features may often be difficult. Other benign and malignant pathologies can also mimic papillary lesions on imaging, and tissue diagnosis is essential. Imaging plays an important role in lesion identification, assessment of extent, tissue sampling, and follow-up. Surgical excision has been recommended for all papillary lesions due to an increased incidence of high-risk lesions and neoplasia even with percutaneous, biopsy-proven benign papillomas. This review looks at papillary breast lesions from the radiologists' standpoint and discusses the clinical, imaging, and pathological features of these lesions, as well as the role of imaging in their evaluation.

Benign oesophageal diseases: a review of the CT findings

A variety of benign conditions can affect the oesophagus, both primarily and secondarily. Traditionally fluoroscopic oesophagography and endoscopy have formed the mainstay for investigating oesophageal diseases. Increasing use of cross-sectional imaging means that many such diseases and their thoracic complications are now detected at computed tomography (CT), which is very often the first imaging tool. This review describes the CT manifestations of a range of benign oesophageal conditions and discusses the role of CT in their evaluation.

Bilateral multicystic dysplastic kidneys -- an unusual case diagnosed on antenatal ultrasound

While it is true that unilateral multicystic dysplastic kidney is commonly associated with abnormalities in the contralateral kidney, it is not widely known that the condition may be bilateral. The recognition of the bilaterality of the multicystic dysplastic kidney by ultrasound has grave prognostic implications as this condition is uniformly fatal. High resolution ultrasound examination of the foetus enables early detection of congenital malformations of the urinary tract. This information is of value in determining foetalprognosis, in deciding the method of delivery, and in alerting the paediatricians to supportive and corrective measures which may need to be taken in postnatal period.

A re-look at an old disease: a multimodality review on gout

Gout, the most common form of microcrystalline arthropathy has always interested radiologists. The diagnosis of gout is primarily based on clinical and laboratory findings; however, it has well known and characteristic radiographic manifestations. Radiographs remain the examination of choice in the diagnosis of joint involvement. Plain radiographs are less sensitive to early changes in gout than other imaging techniques. Recently, magnetic resonance imaging (MRI), ultrasound (US), and computed tomography (CT) have demonstrated an increasing role in early diagnosis of gouty arthritis, for assessing the extent of soft-tissue involvement and as problem-solving tools for diagnostically difficult cases. Cross-sectional imaging can also be used for guiding needle aspirations in patients with an acute attack of gout, which may simulate an infective process clinically. This pictorial review illustrates the main imaging features of gout on radiographs, MRI, CT, and ultrasound with the aim of helping the radiologist to make a confident diagnosis in radiographically typical cases and to serve as a problem-solving tool in cases that present a diagnostic dilemma.

Congenital infiltrating lipomatosis of the face and neck

Congenital infiltrating lipomatosis is a rare clinicopathologic entity characterized by infiltrating lipomatous tumors which, although of benign nature, have a tendency to recur after surgery. This has a predilection for the extremities and the trunk and is seen as overgrowth of soft tissue and bone. It rarely affects the face and neck. We describe two cases of congenital infiltrating lipomatosis of face and neck depicting the bone and soft tissue changes seen on computed tomography, along with a review of the literature on the subject.