Development of a natural language processing algorithm for the detection of spinal metastasis based on magnetic resonance imaging reports.

Background: Metastasis to the spinal column is a common complication of malignancy, potentially causing pain and neurologic injury. An automated system to identify and refer patients with spinal metastases can help overcome barriers to timely treatment. We describe the training, optimization and validation of a natural language processing algorithm to identify the presence of vertebral metastasis and metastatic epidural cord compression (MECC) from radiology reports of spinal MRIs. Methods: Reports from patients with spine MRI studies performed between January 1, 2008 and April 14, 2019 were reviewed by a team of radiologists to assess for the presence of cancer and generate a labeled dataset for model training. Using regular expression, impression sections were extracted from the reports and converted to all lower-case letters with all nonalphabetic characters removed. The reports were then tokenized and vectorized using the doc2vec algorithm. These were then used to train a neural network to predict the likelihood of spinal tumor or MECC. For each report, the model provided a number from 0 to 1 corresponding to its impression. We then obtained 111 MRI reports from outside the test set, 92 manually labeled negative and 19 with MECC to test the model's performance. Results: About 37,579 radiology reports were reviewed. About 36,676 were labeled negative, and 903 with MECC. We chose a cutoff of 0.02 as a positive result to optimize for a low false negative rate. At this threshold we found a 100% sensitivity rate with a low false positive rate of 2.2%. Conclusions: The NLP model described predicts the presence of spinal tumor and MECC in spine MRI reports with high accuracy. We plan to implement the algorithm into our EMR to allow for faster referral of these patients to appropriate specialists, allowing for reduced morbidity and increased survival.

Feasibility study of a novel portable digital radiography system modified for fluoroscopy in the neonatal intensive care unit.

BACKGROUND: A portable radiographic system capable of fluoroscopic imaging in the neonatal intensive care unit (NICU) potentially benefits critically ill neonates by eliminating the need to transport them to a fluoroscopy suite. OBJECTIVE: To evaluate whether a portable bedside fluoroscopy system in the NICU can deliver comparable image quality at a similar dose rate to a standard system in a fluoroscopy suite. MATERIALS AND METHODS: In phase A, 20 patients <3 years of age and scheduled to undergo upper gastrointestinal series (upper GI) or voiding cystourethrograms (VCUG) in the radiology fluoroscopy suite were recruited to evaluate a portable fluoroscopic unit. A modified portable radiographic system with a cassette-sized detector and an in-room fluoroscopy system were sequentially used in the same examination. Four radiologists compared the image quality of 20 images from each system using the Radlex score (1-4) for five image quality attributes. The radiation dose rates for the portable and in-suite systems were collected. In phase B, fluoroscopy studies were performed in 5 neonates in the NICU and compared to the 20 previous neonatal studies performed in the department. Clinical workflow, examination time, fluoroscopy time, scattered radiation dose and patient radiation dose were evaluated. RESULTS: In phase A, average dose rates for in-room and portable systems were equivalent, (0.322 mGy/min and 0.320 mGy/min, respectively). Reader-averaged Radlex scores for in-room and portable systems were statistically significantly greater (P<0.05) for all attributes on the portable system except for image contrast. In phase B, scattered radiation from the average fluoroscopy time (26 s) was equivalent to the scattered radiation of 2.6 portable neonatal chest radiographs. Procedure time and diagnostic quality were deemed equivalent. The average dose rate in the NICU with the portable system was 0.21 mGy/min compared to 0.29 mGy/min for the in-room system. CONCLUSION: The portable fluoroscopy unit is capable of providing comparable image quality at equivalent dose levels to an in-room system for neonates with minimal risks to the staff and other patients in the NICU.

Cytotoxic lesion of the splenium of the corpus callosum in a patient with EVALI.

Since first recognized in July 2019, numerous cases of a lung illness associated with electronic-cigarette use or vaping particularly tetrahydrocannabinol in adolescents and adults have been reported. As of January 2020, the Center for Disease Control has reported over 2500 cases of electronic-cigarette or vaping product use-associated lung injury (EVALI), including 60 deaths. Affected patients most commonly present with respiratory or gastrointestinal complaints although neurological symptoms including headache, confusion and lethargy have been reported. We present a new as yet unpublished finding in the brain of a previously healthy teenage boy with EVALI. Brain imaging may be warranted in this patient population.

Imaging findings in systemic childhood diseases presenting with dermatologic manifestations.

PURPOSE: Many childhood diseases often present with skin abnormalities with which radiologists are largely unfamiliar. Knowledge of associated dermatologic manifestations may aid the radiologist in confirming the diagnosis and recommending targeted imaging of affected organs. METHODS: We review the imaging findings in childhood diseases associated with dermatologic manifestations. FINDINGS: Diseases include dermatologic findings which herald underlying malignancy (Neuroblastoma, leukemia/lymphoma, Langerhans cell histiocytosis),are associated with risk of malignancy (Epidermolysis Bullosa, basal cell nevus syndrome, Cowden's syndrome, Tuberous Sclerosis),or indicate a systemic inflammatory/immune disorder (Kawasaki's disease, Henoch Schonlein Purpura, systemic lupus erythematosus, scleroderma, sarcoidosis, dermatomyositis and immune thrombocytopenic purpura). CONCLUSION: Familiarity with pertinent findings in childhood diseases presenting with dermatologic manifestations in childhood diseases aids the radiologist in confirming the diagnosis and guiding imaging workup.

Scoring system for differentiating perforated and non-perforated pediatric appendicitis.

OBJECTIVES: Appendicitis is the most common indication for emergency pediatric surgery and its most significant complication is perforation. Perforated appendicitis (PA) may be managed conservatively, whereas non-perforated appendicitis (NP) is managed surgically. Recent studies have shown that ultrasound (US) is effective for differentiating between PA and NP, and does not expose pediatric patients to ionizing radiation. The purpose of this study is to enhance the accuracy of differentiation with a novel scoring system based on clinical, laboratory, and US findings. METHODS: This retrospective study included 243 patients aged 2-17 years who presented between 2006 and 2013 with surgically proven appendicitis, of whom 60 had perforation. Clinical and laboratory data were collected and US images evaluated by a pediatric radiologist. To create the scoring system, point values were assigned to each parameter. A randomly selected training sample of 137 subjects was used to create a scoring prediction model. The model was tested on the remaining 106 patients. RESULTS: Scores of ≥6, ≥11, and ≥15 yielded specificities of 64, 91, and 99%, and sensitivities of 96, 61, and 29%, respectively (p < 0.001). CONCLUSIONS: We have designed a scoring system incorporating clinical, laboratory, and sonographic findings which can differentiate PA from NP with high specificity.

Evaluation of an initiative to reduce radiation exposure from CT to children in a non-pediatric-focused facility.

We would like to share our experience of reducing pediatric radiation exposure. Much of the recent literature regarding successes of reducing radiation exposure has come from dedicated children's hospitals. Nonetheless, over the past two decades, there has been a considerable increase in CT imaging of children in the USA, predominantly in non-pediatric-focused facilities where the majority of children are treated. In our institution, two general hospitals with limited pediatric services, a dedicated initiative intended to reduce children's exposure to CT radiation was started by pediatric radiologists in 2005. The initiative addressed multiple issues including eliminating multiphase studies, decreasing inappropriate scans, educating referring providers, training residents and technologists, replacing CT with ultrasound or MRI, and ensuring availability of pediatric radiologists for consultation. During the study period, the total number of CT scans decreased by 24 %. When accounting for the number of scans per visit to the emergency department (ED), the numbers of abdominal and head CT scans decreased by 37.2 and 35.2 %, respectively. For abdominal scans, the average number of phases per scan decreased from 1.70 to 1.04. Upon surveying the pediatric ED staff, it was revealed that the most influential factors on ordering of scans were daily communication with pediatric radiologists, followed by journal articles and lectures by pediatric radiologists. We concluded that a non-pediatric-focused facility can achieve dramatic reduction in CT radiation exposure to children; however, this is most effectively achieved through a dedicated, multidisciplinary process led by pediatric radiologists.

Ultrasound for differentiation between perforated and nonperforated appendicitis in pediatric patients.

OBJECTIVE: Acute appendicitis is the most common condition requiring emergency surgery in children. Differentiation of perforated from nonperforated appendicitis is important because perforated appendicitis may initially be managed conservatively whereas nonperforated appendicitis requires immediate surgical intervention. CT has been proved effective in identifying appendiceal perforation. The purpose of this study was to determine whether perforated and nonperforated appendicitis in children can be similarly differentiated with ultrasound. MATERIALS AND METHODS: This retrospective study included 161 consecutively registered children from two centers who had acute appendicitis and had undergone ultra-sound and appendectomy. Ultrasound images were reviewed for appendiceal size, appearance of the appendiceal wall, changes in periappendiceal fat, and presence of free fluid, abscess, or appendicolith. The surgical report served as the reference standard for determining whether perforation was present. The specificity and sensitivity of each ultrasound finding were determined, and binary models were generated. RESULTS: The patients included were 94 boys and 67 girls (age range, 1-20 years; mean, 11 ± 4.4 [SD] years) The appendiceal perforation rate was significantly higher in children younger than 8 years (62.5%) compared with older children (29.5%). Sonographic findings associated with perforation included abscess (sensitivity, 36.2%; specificity, 99%), loss of the echogenic submucosal layer of the appendix in a child younger than 8 years (sensitivity, 100%; specificity, 72.7%), and presence of an appendicolith in a child younger than 8 years (sensitivity, 68.4%; specificity, 91.7%). CONCLUSION: Ultrasound is effective for differentiation of perforated from nonperforated appendicitis in children.

Pott's puffy tumor, intracranial, and orbital complications as the initial presentation of sinusitis in healthy adolescents, a case series.

Pott's puffy tumor (PPT) and intracranial complications of sinusitis are considered rare in the post-antibiotic era. We present a series of pediatric patients who were diagnosed with these complications. The purpose of this study is to discuss the clinical presentations and imaging findings in order to increase awareness of pediatricians and radiologists. A computer search identified nine pediatric patients who were admitted to our hospital between October 2006 and July 2009. The patients' charts and imaging studies were reviewed. There were eight males (11-16 years old) and one female (9 years old). All patients presented with significant headache and fever. Only two patients had sinusitis symptoms. CT and MRI revealed sinusitis in seven patients. These patients were diagnosed with epidural abscess (six), subdural empyema (one), PPT (five), focal meningitis (six), and orbital cellulitis (four). Two patients did not have sinusitis on the day of admission, but were treated recently for sinusitis. These two patients presented with frontal epidural abscess, PPT, thrombosis of the superior sagittal sinus, and frontal bone osteomyelitis. All patients were treated with a multidisciplinary approach and had good outcome with no residual neurologic deficits. Complications of sinusitis are not rare in the studied population. The presentation may be misleading. Therefore, a high index of suspicion is needed, particularly in adolescent males who present with significant headache and fever. Evaluation with contrast-enhanced CT of the brain, sinuses, and orbits should be done first, followed by MRI examinations.

Interloop fluid in intussusception: what is its significance?

BACKGROUND: Sonography has been used to predict pneumatic reduction outcome in children with intussusception. OBJECTIVE: To assess the prognostic significance of fluid between the intussusceptum and intussuscepiens with respect to reduction outcome, lead point or necrosis. MATERIALS AND METHODS: Sonograms of children with a discharge diagnosis of intussusception from four institutions were reviewed for interloop fluid and correlated with results of pneumatic reduction and surgical/pathological findings when available. Maximal dimension of interloop fluid on a transverse image and fluid complexity were evaluated. RESULTS: Of 166 cases, 36 (21.7%) had interloop fluid. Pneumatic reduction was successful in 21 (58.3%) with fluid and 113 (87.6%) without. The average largest fluid dimension was 8.7 mm (range 5 mm-19 mm, median 8 mm) in cases with successful reduction and 12.8 mm (range 4 mm-26 mm, median 12.5 mm) in unsuccessful reduction (p < 0.05). Fluid dimension equal to or greater than 9 mm correlated with failed reduction (p < 0.0001;odds ratio 13:1). In 36 cases with interloop fluid that required surgery, there were four lead points and three necrosis. In cases without fluid with surgical reduction, there was one lead point and one necrosis. Interloop fluid correlated with lead point (p < 0.04) or necrosis (p < 0.03). Its significance increased with larger amounts of fluid (p < 0.0001). Patient age/fluid complexity did not correlate with reduction outcome (p = 0.9). CONCLUSION: Interloop fluid was associated with increased failure of pneumatic reduction and increased likelihood of lead point or necrosis, particularly when the maximum dimension exceeded 9 mm.