Automated abnormality classification of chest radiographs using deep convolutional neural networks.

As one of the most ubiquitous diagnostic imaging tests in medical practice, chest radiography requires timely reporting of potential findings and diagnosis of diseases in the images. Automated, fast, and reliable detection of diseases based on chest radiography is a critical step in radiology workflow. In this work, we developed and evaluated various deep convolutional neural networks (CNN) for differentiating between normal and abnormal frontal chest radiographs, in order to help alert radiologists and clinicians of potential abnormal findings as a means of work list triaging and reporting prioritization. A CNN-based model achieved an AUC of 0.9824 ± 0.0043 (with an accuracy of 94.64 ± 0.45%, a sensitivity of 96.50 ± 0.36% and a specificity of 92.86 ± 0.48%) for normal versus abnormal chest radiograph classification. The CNN model obtained an AUC of 0.9804 ± 0.0032 (with an accuracy of 94.71 ± 0.32%, a sensitivity of 92.20 ± 0.34% and a specificity of 96.34 ± 0.31%) for normal versus lung opacity classification. Classification performance on the external dataset showed that the CNN model is likely to be highly generalizable, with an AUC of 0.9444 ± 0.0029. The CNN model pre-trained on cohorts of adult patients and fine-tuned on pediatric patients achieved an AUC of 0.9851 ± 0.0046 for normal versus pneumonia classification. Pretraining with natural images demonstrates benefit for a moderate-sized training image set of about 8500 images. The remarkable performance in diagnostic accuracy observed in this study shows that deep CNNs can accurately and effectively differentiate normal and abnormal chest radiographs, thereby providing potential benefits to radiology workflow and patient care.

US of the Peripheral Nerves of the Upper Extremity: A Landmark Approach.

Ultrasonography (US) has become a first-line modality for the evaluation of the peripheral nerves of the upper extremity. The benefits of US over magnetic resonance (MR) imaging include higher soft-tissue resolution, cost effectiveness, portability, real-time and dynamic imaging, and the ability to scan an entire extremity quickly and efficiently. US can be performed on patients who are not eligible for MR imaging. Metallic implant artifacts are usually not problematic. US has been shown to have equal specificity and greater sensitivity than MR imaging in the evaluation of peripheral nerves. Any abnormal findings can be easily compared with the contralateral side. The published literature has shown that US has demonstrated clinical utility in patients with suspected peripheral nerve disease by guiding diagnostic and therapeutic decisions as well as by confirming electrodiagnostic findings. Common indications for upper extremity peripheral nerve US are the evaluation for injury due to penetrating trauma, entrapment by scar tissue, and tumor. US of the upper extremity is most commonly performed to evaluate carpal and cubital tunnel syndrome. It is important for the radiologist or sonographer to have a detailed knowledge of anatomy and specific anatomic landmarks for each nerve to efficiently and accurately perform an examination. The goal of this article is to introduce readers to the basics of US of the peripheral nerves of the upper extremity with a focus on the median, ulnar, and radial nerves. Common sites of disease and the location of important anatomic landmarks will be reviewed.

US of the Peripheral Nerves of the Lower Extremity: A Landmark Approach.

Ultrasonography (US) is commonly used to assess the peripheral nerves of the lower extremity because of its many advantages over magnetic resonance (MR) imaging. The most obvious advantages over MR imaging are superior soft-tissue resolution, low cost, portability, lack of magnetic susceptibility artifact, and the ability to image patients who cannot undergo MR imaging. US has been shown to have equal specificity and greater sensitivity than MR imaging in the evaluation of peripheral nerves. Additional benefits are the capability of real-time and dynamic imaging, and the ability to scan an entire extremity quickly without the need for a patient to lie motionless for long periods of time, as with MR imaging. Any abnormal findings can be easily compared against the contralateral side. Published literature has shown that US has clinical utility in patients suspected of having peripheral nerve disease: US can be used to guide diagnostic and therapeutic decisions, as well as help confirm electrodiagnostic findings. Common indications for lower extremity peripheral nerve US are the evaluation for injury due to penetrating trauma, entrapment by scar tissue, or tumor. To confidently perform US of the peripheral nerves of the lower extremity, it is important to gain a thorough knowledge of anatomic landmarks and the course of each nerve. Readers who may not be familiar with US will be introduced to the basics of scanning the peripheral nerves of the lower extremity. Important anatomic landmarks and common sites of injury and entrapment will be reviewed.

Role of Sonography in Clinically Occult Femoral Hernias.

OBJECTIVES: The purpose of this article is to evaluate the diagnostic accuracy of sonography in clinically occult femoral hernias and to describe our sonographic technique. METHODS: The clinical and imaging data for 93 outpatients referred by general surgeons, all of whom underwent sonographic evaluation and surgery, were reviewed retrospectively. Of these, 55 patients who underwent surgical exploration for groin hernias within 3 months of sonography and met all inclusion criteria were included in the study. The sonographic technique involves using the pubic tubercle as an osseous landmark to identify and appropriately visualize the femoral canal. The Valsalva maneuver is then used to differentiate the movement of normal fat (a potential pitfall) from true herniation in the femoral canal. Surgical findings were used as the reference standard by which sonographic results were judged. Two-by-two contingency tables were used to calculate the sensitivity, specificity, positive predictive value, and negative predictive value. RESULTS: In these 55 patients, surgery revealed 15 femoral hernias. Eight femoral hernias occurred in women, and 7 occurred in men. For diagnosing femoral hernias, sonography demonstrated sensitivity of 80%, specificity of 88%, a positive predictive value of 71%, and a negative predictive value of 92%. True-positive cases of femoral hernias have a sonographic appearance of a hypoechoic sac with speckled internal echoes. When examining during the Valsalva maneuver, a femoral hernia passes deep to the inguinal ligament, expands the femoral canal, displacing the normal canal fat, and effaces the femoral vein. CONCLUSIONS: Sonography can exclude femoral hernias with high confidence in light of its exceptional negative predictive value. With attention to technique and imaging criteria, the diagnostic accuracy of sonography can be enhanced.

Ultrasound of the Groin: Techniques, Pathology, and Pitfalls.

OBJECTIVE: In a patient with symptoms referable to the groin, there can be a number of causes to consider and at times the cause of the symptoms is multifactorial. Although ultrasound can be effective in the evaluation of the groin, the depth and complexity of the anatomy can be problematic. A protocol-driven approach for ultrasound evaluation of the groin will help to ensure an accurate and comprehensive evaluation. This article summarizes the ultrasound technique and protocol for evaluation of the groin to include evaluation of the hip joint, anterior hip musculature, the iliopsoas bursa, the inguinal lymph nodes, the pubic symphyseal region, and the inguinal region for hernias. Common pathologic conditions and pitfalls related to ultrasound evaluation of the groin will be reviewed. CONCLUSION: There are many potential pathologic conditions in a patient with groin symptoms. Because symptoms may be multifactorial and history may be ambiguous or misleading, a protocol-driven evaluation with ultrasound is recommended. The hip joint is evaluated for effusion, synovitis, and labral abnormalities. The muscles and tendons, including the common aponeurosis at the pubic symphysis, are evaluated for tendinosis and tears. Dynamic evaluation should be considered to assess for snapping hip syndrome. Iliopsoas bursal distention and lymph node enlargement are other considerations. Last, inguinal region hernias must be evaluated during the Valsalva maneuver and documented in two orthogonal planes to avoid several important diagnostic pitfalls.

Predicting Birth-Related Levator Ani Tear Severity in Primiparous Women: Evaluating Maternal Recovery from Labor and Delivery (EMRLD Study).

OBJECTIVE: To determine which maternal characteristics or birth events independently predict severity of levator ani muscle (LA) tears at first vaginal birth in a longitudinal/observational investigation in a tertiary care hospital. SAMPLE: Ninety primiparas with at least one at risk for LA tear inclusion factor at vaginal birth: maternal age ≥ 33 years, second stage ≥ 150 minutes, macrosomia, instrumented delivery, and/or anal sphincter laceration were studied. METHODS: Magnetic Resonance Imaging (MRI) was obtained early postpartum (mean ± sd 48.9 ± 21.6 days) to identify LA tear. Severity of LA muscle fiber loss was graded on an ordinal scale of: "0" as no loss, "1" as <50% unilateral loss, "2" as ≥50% unilateral or <50% bilateral loss, and "3" as ≥50% bilateral loss. Data were analyzed using proportional odds modeling. Inclusion factors were explored as predictors of LA tear severity and at analysis episiotomy, time spent actively pushing, epidural, and oxytocin were also considered. The main outcome measures of interest included grading of severity of LA muscle fiber loss on an ordinal scale. RESULTS: Respective counts/percentages of women within each 0 thru 3 ordered category of LA tear severity were: "0" = 58(64%), "1" = 9(10%), "2" = 15(17%), and "3" = 8(9%). Estimates and 95% CI for significant demographic or obstetric univariate predictors of LA tear severity level were age, OR 1.093 (CI 1.012 - 1.180), p = 0.023; and time spent in active pushing, OR 1.089 (CI 1.005 - 1.180), p = 0.038. The other factors considered were not significant. There were too few women with forceps deliveries to analyze. CONCLUSION: In our enriched sample of primiparous women, 26% showed a significant LA tear. Maternal age and time spent actively pushing independently predict LA tear severity.

Imaging appearances in gout.

Gout is an ancient disease. Last decade has brought about significant advancement in imaging technology and real scientific growth in the understanding of the pathophysiology of gout, leading to the availability of multiple effective noninvasive diagnostic imaging options for gout and treatment options fighting inflammation and controlling urate levels. Despite this, gout is still being sub-optimally treated, often by nonspecialists. Increased awareness of optimal treatment options and an increasing role of ultrasound and dual energy computed tomography (DECT) in the diagnosis and management of gout are expected to transform the management of gout and limit its morbidity. DECT gives an accurate assessment of the distribution of the deposited monosodium urate (MSU) crystals in gout and quantifies them. The presence of a combination of the ultrasound findings of an effusion, tophus, erosion and the double contour sign in conjunction with clinical presentation may be able to obviate the need for intervention and joint aspiration in a certain case population for the diagnosis of gout. The purpose of this paper is to review imaging appearances of gout and its clinical applications.

Teaching medical management and operations engineering for systems-based practice to radiology residents.

RATIONALE AND OBJECTIVES: To better prepare radiology residents for providing care within the context of the larger health care system, this study evaluated the feasibility and impact of a curriculum to enhance radiology residents' understanding and ability to apply concepts from medical management and industrial and operational engineering to systems-based practice problems in radiology practice. MATERIALS AND METHODS: A multiprofessional team including radiology, medical education, and industrial and operational engineering professionals collaborated in developing a seven-module curriculum, including didactic lectures, interactive large-group analysis, and small-group discussions with case-based radiology examples, which illustrated real-life management issues and the roles physicians held. Residents and faculty participated in topic selection. Pre- and post-instruction formative assessments were administered, and results were shared with residents during teaching sessions. RESULTS: Attendance and participation in case-based scenario resolutions indicate the feasibility and impact of the interactive curriculum on residents' interest and ability to apply curricular concepts to systems-based practice in radiology. Paired t test analyses (P < .05) and effect sizes showed residents significantly increased their knowledge and ability to apply concepts to systems-based practice issues in radiology. CONCLUSIONS: Our iterative curriculum development and implementation process demonstrated need and support for a multiprofessional team approach to teach management and operational engineering concepts. Curriculum topics are congruent with Accreditation Council for Graduate Medical Education requirements for systems-based practice. The case-based curriculum using a mixed educational format of didactic lectures and small-group discussion and problem analysis could be adopted for other radiology programs, for both residents and continuing medical education applications.

Extra-abdominal desmoid tumor with osseous involvement.

Deep fibromatoses, or desmoid tumors, arise from connective tissue. Imaging can frequently suggest the diagnosis of these aggressive, benign neoplasms. Cross-sectional imaging commonly demonstrates an enhancing solid mass that resembles scar tissue, typically without osseous involvement. We report a case of an extra-abdominal desmoid tumor involving the teres minor muscle in a symptomatic 42-year-old woman with unusual features of medullary involvement and negative nuclear beta-catenin staining.

Partnering with Engineers to Identify and Empirically Evaluate Delays in Magnetic Resonance Imaging Laying the Foundations for Quality Improvement and System-based Practice in Radiology.

RATIONALE AND OBJECTIVES: The aim of this study was to evaluate the feasibility of partnering with engineering students and critically examining the merit of the problem identification and analyses students generated in identifying sources impeding effective turnaround in a large university department of diagnostic radiology. Turnaround involves the time and activities beginning when a patient enters the magnetic resonance scanner room until the patient leaves, minus the time the scanner is conducting the protocol. MATERIALS AND METHODS: A prospective observational study was conducted, in which four senior undergraduate industrial and operations engineering students interviewed magnetic resonance staff members and observed all shifts. On the basis of 150 hours of observation, the engineering students identified 11 process steps (eg, changing coils). They charted machine use for all shifts, providing a breakdown of turnaround time between appropriate process and non-value-added time. To evaluate the processes occurring in the scanning room, the students used a work-sampling schedule in which a beeper sounded 2.5 times per hour, signaling the technologist to identify which of 11 process steps was occurring. This generated 2147 random observations over a 3-week period. RESULTS: The breakdown of machine use over 105 individual studies showed that non-value-added time accounted for 62% of turnaround time. Analysis of 2147 random samples of work showed that scanners were empty and waiting for patients 15% of the total time. Analyses showed that poor communication delayed the arrival of patients and that no one had responsibility for communicating when scanning was done. CONCLUSIONS: Engineering students used rigorous study design and sampling methods to conduct interviews and observations. This led to data-driven definition of problems and potential solutions to guide systems-based improvement.

Groin pain beyond the hip: how anatomy predisposes to injury as visualized by musculoskeletal ultrasound and MRI.

OBJECTIVE: The purpose of this article is to show ultrasound and MRI examples of the normal anatomic structures and their resulting modifications from trauma and disease. CONCLUSION: Although groin pain from hip pathology is well recognized, lower anterior abdominal wall and anterior pelvis structures can be interrelated sources of pain.

Patients' perception of care during image-guided breast biopsy in a rural community breast center: communication matters.

This retrospective study surveyed women about their experience with image-guided breast biopsies in a rural cancer center. Our study objectives were to determine: women's perception of their emotional and physical comfort during the biopsy; the clarity and acceptability of the communication about the biopsy, as well as the methods used to provide the results of her biopsy and treatment options; and the overall impressions of the technical and personal care she received. A single mailed survey of 500 biopsy patients. The response rate was 43% (n = 215). Prior to their biopsy, 22% had been concerned that the biopsy would be painful and 43% were concerned that the biopsy might show cancer. Almost all women rated the communication about the biopsy procedure as understandable (99%) and accurately reflecting their experience (99%.). Most (77%) patients characterized the biopsy as producing minimal discomfort, although 5% disagreed. Most patients (98%) characterized their physician and technologists as caring about their emotional and physical comfort. Most patients felt that the typical 1-day wait to receive the results of their biopsy was reasonable (90%) and that the use of the phone to convey the results was acceptable (90%). On comparing patients with a diagnosis of cancer to those without, there was no difference in the level of satisfaction with the use of the phone to communicate biopsy results. This study offers encouragement that communication practices can alleviate the anxiety of women undergoing image-guided breast cancer biopsies.

Pelvic structure and function at 1 month compared to 7 months by dynamic magnetic resonance after vaginal birth.

OBJECTIVE: We sought to determine whether changes exist in location and movement of pelvic floor structures at 1 and 7 months postpartum. STUDY DESIGN: Midsagittal magnetic resonance images from 13 primiparous women with birth events associated with levator ani damage at early ( approximately 1 month) and late ( approximately 7 months) postpartum time points were analyzed. Pelvic floor structure locations at rest and displacements from rest to maximum Kegel and Valsalva were determined. Urogenital and levator hiatus diameters were measured as well. RESULTS: The perineal body was 7.1 mm and anal verge 7.9 mm higher at 7 months postpartum (P = .003). Both the urogenital and levator hiatus diameters were smaller at 7 months (P < .05). Displacement during Kegel and Valsalva was similar between the 2 time points. CONCLUSION: Resting locations of the perineal body and anal verge are higher at 7 months postpartum, but the amount of movement during Kegel or Valsalva does not change.

Anatomy of the perineal membrane as seen in magnetic resonance images of nulliparous women.

OBJECTIVE: Recent cadaver research demonstrates the perineal membrane's ventral and dorsal portions and close relationship to the levator ani muscle. This study seeks to show these relationships in women by magnetic resonance (MR) images. STUDY DESIGN: The subjects were 20 asymptomatic nulliparous women with normal pelvic examinations. MR images were acquired in multiple planes. Anatomical relationships from cadaver studies were examined in these planes. RESULTS: In the coronal plane the ventral perineal membrane forms an interconnected complex with the compressor urethrae, vestibular bulb, and levator ani. The dorsal part connects the levator ani and vaginal side wall via a distinct band to the ischiopubic ramus. In the sagittal plane the parallel position of perineal membrane and levator ani are seen. CONCLUSION: The perineal membrane's anatomical features can be seen in women with MR. The close relationship between the perineal membrane and levator ani is evident.

MRI interpretation proficiency of musculoskeletal fellows in training.

RATIONALE AND OBJECTIVES: The aim of this study was to evaluate the magnetic resonance imaging (MRI) interpretation proficiency of musculoskeletal fellows in training. MATERIALS AND METHODS: Between July 2003 and June 2007, 14 musculoskeletal fellows were independently tested with 20 MRI studies of the knee and shoulder at four separate time points during their fellowship years. Trends in true-positive and false-positive interpretation results were evaluated. Fellows who completed their residencies at the fellowship institution (internal fellows) were compared with those from other residencies (external fellows). RESULTS: There was a significant improvement in proficiency between the initial and final (9-month) evaluations (P < .0001). At the initial evaluation, there was a mean of 52.8% (41.7 of 79) true-positive results (range, 32-51); at 9 months, there was a mean of 71.0% (56.1 of 79; range, 40-72). The number of false-positive results also declined during this time period from a mean of 8.1 (range, 2-13) at initial evaluation to 4.7 (range, 2-8) at 9 months (P < .001). External fellows had more incorrect diagnoses initially but showed greater improvement than internal fellows at 9 months. CONCLUSION: Fellows continued to improve their MRI interpretation skills throughout the first 9 months of their fellowships. External fellows were slightly less proficient at the start of their fellowships but slightly more proficient at 9 months compared to internal fellows.

Abdominal wall hernia mesh repair: sonography of mesh and common complications.

OBJECTIVE: The purposes of this study were (1) to review the sonographic in vitro and in vivo appearances of mesh for surgical repair of abdominal wall hernias, (2) to describe sonographic techniques and discuss the limitations of sonography in evaluation of mesh hernia repair, and (3) to illustrate common complications after mesh repair shown with sonography. METHODS: We identified interesting cases from the musculoskeletal sonographic database as well as from the teaching files of the authors, with surgical or other cross-sectional imaging corroboration. RESULTS: A compilation of the sonographic appearances of mesh used for anterior abdominal wall and inguinal hernia repair and complications diagnosable by sonography is presented. CONCLUSIONS: Sonography can be effective for evaluation of mesh and complications after mesh repair of anterior abdominal wall and inguinal hernias.