Original research: utilization of a convolutional neural network for automated detection of lytic spinal lesions on body CTs.

OBJECTIVE: To develop, train, and test a convolutional neural network (CNN) for detection of spinal lytic lesions in chest, abdomen, and pelvis CT scans. MATERIALS AND METHODS: Cases of malignant spinal lytic lesions in CT scans were identified. Images were manually segmented for the following classes: (i) lesion, (ii) normal bone, (iii) background. If more than one lesion was on a single slice, all lesions were segmented. Images were stored as 128×128 pixel grayscale, with 10% segregated for testing. The training pipeline of the dataset included histogram equalization and data augmentation. A model was trained on Keras/Tensorflow using an 80/20 training/validation split, based on U-Net architecture. Additional testing of the model was performed on 1106 images of healthy controls. Global sensitivity measured detection of any lesion on a single image. Local sensitivity and positive predictive value (PPV) measured detection of all lesions on an image. Global specificity measured false positive rate in non-pathologic bone. RESULTS: Six hundred images were obtained for model creation. The training set consisted of 540 images, which was augmented to 20,000. The test set consisted of 60 images. Model training was performed in triplicate. Mean Dice scores were 0.61 for lytic lesion, 0.95 for normal bone, and 0.99 for background. Mean global sensitivity was 90.6%, local sensitivity was 74.0%, local PPV was 78.3%, and global specificity was 63.3%. At least one false positive lesion was noted in 28.8-44.9% of control images. CONCLUSION: A task-trained CNN showed good sensitivity in detecting spinal lytic lesions in axial CT images.

Automated detection and segmentation of sclerotic spinal lesions on body CTs using a deep convolutional neural network.

PURPOSE: To develop a deep convolutional neural network capable of detecting spinal sclerotic metastases on body CTs. MATERIALS AND METHODS: Our study was IRB-approved and HIPAA-compliant. Cases of confirmed sclerotic bone metastases in chest, abdomen, and pelvis CTs were identified. Images were manually segmented for 3 classes: background, normal bone, and sclerotic lesion(s). If multiple lesions were present on a slice, all lesions were segmented. A total of 600 images were obtained, with a 90/10 training/testing split. Images were stored as 128 × 128 pixel grayscale and the training dataset underwent a processing pipeline of histogram equalization and data augmentation. We trained our model from scratch on Keras/TensorFlow using an 80/20 training/validation split and a U-Net architecture (64 batch size, 100 epochs, dropout 0.25, initial learning rate 0.0001, sigmoid activation). We also tested our model's true negative and false positive rate with 1104 non-pathologic images. Global sensitivity measured model detection of any lesion on a single image, local sensitivity and positive predictive value (PPV) measured model detection of each lesion on a given image, and local specificity measured the false positive rate in non-pathologic bone. RESULTS: Dice scores were 0.83 for lesion, 0.96 for non-pathologic bone, and 0.99 for background. Global sensitivity was 95% (57/60), local sensitivity was 92% (89/97), local PPV was 97% (89/92), and local specificity was 87% (958/1104). CONCLUSION: A deep convolutional neural network has the potential to assist in detecting sclerotic spinal metastases.

A measuring technique for intra-osseous pressure.

OBJECTIVE: Pulsatile intra-osseous pressures result in bone remodeling, and therefore may affect lesion growth and response to treatment. However, there is no known method used to measure intra-osseous pressures. The purpose of this study is to describe a novel image-guided technique for measuring intra-osseous pressures. MATERIALS AND METHODS: This study was IRB-approved and HIPAA compliant. Written informed consent was obtained. Intra-osseous pressure measurements were performed during a CT-guided bone marrow biopsy in eight patients (6 male, 2 female) with mean age 66 ± 13 years (median 72, range 45-87) and suspected or known bone marrow disease. Bone marrow pressure measurements were obtained connecting the biopsy needle to a dedicated monitor using a standard arterial line setup. Monitor data was collected at 5-s intervals in order to record continuous pressure measurements for 2 min. RESULTS: Pressure measurements were successfully performed in all 8 patients. The mean bone marrow pressures were 36.8 ± 7.2 mmHg (median 37.7, range 24.7-47.4). The peak and trough pressures varied by 11%, and the standard deviation of mean pressure measurement varied by 18%. Our findings for marrow pressure measures most closely approximate the pressure profile of the venous system. CONCLUSION: We describe a novel and minimally invasive technique able to provide functional data of bone marrow. This technique has the potential to provide insights into normal and diseased bone marrow and may be helpful to evaluate features of cystic and vascular tumors that may be amenable to percutaneous treatments.

MRI and clinical features of acute fungal discitis/osteomyelitis.

OBJECTIVES: To compare imaging and clinical features of fungal and Staphylococcus aureus discitis-osteomyelitis (DO) for patients presenting for CT-guided biopsies. METHODS: Our study was IRB-approved and HIPAA-compliant. A group of 11 fungal DO (FG) with MRI within 7 days of the biopsy and a control group (CG) of 19 Staphylococcus aureus DO were evaluated. Imaging findings (focal vs diffuse paravertebral soft tissue abnormality, partial vs complete involvement of the disc/endplate), biopsy location, pathology, duration of back pain, immune status, history of intravenous drug, history of prior infection, current antibiotic treatment, and history of invasive intervention. Differences were assessed using the Fisher exact test and Kruskal-Wallis test. Naïve Bayes predictive modeling was performed. RESULTS: The most common fungal organisms were Candida species (9/11, 82%). The FG was more likely to have focal soft tissue abnormality (p = 0.040) and partial disc/endplate involvement (p = 0.053). The clinical predictors for fungal DO, in order of importance, back pain for 10 or more weeks, current antibiotic use for 1 week or more, and current intravenous drug use. History of invasive instrumentation within 1 year was more predictive of Staphylococcus aureus DO. CONCLUSION: MRI features (focal partial soft tissue abnormality and partial involvement of the disc/endplate) in combination with clinical features may help to predict fungal species as a causative organism for DO. KEY POINTS: • MRI features of discitis-osteomyelitis (focal partial soft tissue abnormality and partial involvement of the disc/endplate) in combination with clinical features may help to predict fungal species as a causative organism for DO.

CT-guided discitis-osteomyelitis biopsies with negative microbiology: how many days should we wait before repeating the biopsy?

PURPOSE: To determine the number of days to positive CT-guided biopsy sample culture in patients with discitis-osteomyelitis. METHODS: Our study was IRB approved and HIPAA compliant. All CT-guided biopsies performed for acute discitis-osteomyelitis with positive microbiology between 2002 and 2018 were reviewed. Microbiological organism and days to positive biopsy were documented. Mean, median, skew, and standard deviation were calculated. The proportion of positive cultures that become positive after each day has elapsed was also calculated. RESULTS: There were 96 true positive cultures, with 64 (67%) male and 32 (33%) female, ages 57 ± 18 (range 19-87) years. Overall, including all culture results, the mean number of days to positive culture was 2.9 ± 3.5 days. The median number of days was 2, with a positive skew of 2.9. At days 1, 2, 3, 4, and 5, 48%, 68%, 78%, 85%, and 89%, respectively, of biopsy samples had a positive microbiology culture. CONCLUSION: Approximately three-quarters of discitis-osteomyelitis pathogens will be identified by biopsy sample culture by 3 days after CT-guided biopsy. This finding should be considered if planning for a repeat biopsy in the setting of a negative microbiology culture.