Effect of a Deep Learning Framework-Based Computer-Aided Diagnosis System on the Diagnostic Performance of Radiologists in Differentiating between Malignant and Benign Masses on Breast Ultrasonography

OBJECTIVE: To investigate whether a computer-aided diagnosis (CAD) system based on a deep learning framework (deep learning-based CAD) improves the diagnostic performance of radiologists in differentiating between malignant and benign masses on breast ultrasound (US). MATERIALS AND METHODS: B-mode US images were prospectively obtained for 253 breast masses (173 benign, 80 malignant) in 226 consecutive patients. Breast mass US findings were retrospectively analyzed by deep learning-based CAD and four radiologists. In predicting malignancy, the CAD results were dichotomized (possibly benign vs. possibly malignant). The radiologists independently assessed Breast Imaging Reporting and Data System final assessments for two datasets (US images alone or with CAD). For each dataset, the radiologists' final assessments were classified as positive (category 4a or higher) and negative (category 3 or lower). The diagnostic performances of the radiologists for the two datasets (US alone vs. US with CAD) were compared. RESULTS: When the CAD results were added to the US images, the radiologists showed significant improvement in specificity (range of all radiologists for US alone vs. US with CAD: 72.8–92.5% vs. 82.1–93.1%; p < 0.001), accuracy (77.9–88.9% vs. 86.2–90.9%; p = 0.038), and positive predictive value (PPV) (60.2–83.3% vs. 70.4–85.2%; p = 0.001). However, there were no significant changes in sensitivity (81.3–88.8% vs. 86.3–95.0%; p = 0.120) and negative predictive value (91.4–93.5% vs. 92.9–97.3%; p = 0.259). CONCLUSION: Deep learning-based CAD could improve radiologists' diagnostic performance by increasing their specificity, accuracy, and PPV in differentiating between malignant and benign masses on breast US.

Comparison of the diagnostic accuracy of clinically significant prostate cancer based on the PI-RADS: an interobserver study / 中华泌尿外科杂志

Objective To compare the diagnostic accuracy of clinically significant prostate cancer by general radiologist and uroradiology specialist based on the Prostate Imaging Reporting and Data System (PI-RADS).Methods A total of 45 men from Beijing United Family Hospital and Clinics undergoing prostate mpMRI examination and subsequent MRI-targeted biopsy were included in the study.The age of patients was (60.0 ± 8.0) years,the median PSA level was 7.2 ng/ml (1.2-95.8 ng/ml) and the median prostate volume was 45.0 ml (18.3-127.0 ml).The general radiologists from Beijing United Family Hospital and Clinics made the diagnosis according to PI-RADS 2.0.One uroradiology specialist from Beijing Anzhen hospital reviewed all the mpMRIs retrospectively and marked new PI-RADS score based on PI-RADS 2.0.The PI-RADS ≥ 3 lesion was recognized as suspicious of clinically significant prostate cancer.The distribution of PI-RADS score from different doctors and the diagnostic accuracy of clinically significant prostate cancer was compared.Results All the 45 patients underwent MRI-targeted cognitive biopsy and 14 cases of prostate cancer were detected,including 9 cases of clinically significant prostate cancer.There was no significant difference in the distribution of PI-RADS by general radiologist and uroradiology specialist (P =0.064).82.8％ (37/45) and 37.8％ (17/45) patients were diagnosed with PI-RADS ≥ 3 by general radiologist and uroradiology specialist respectively.The interobserver agreement was only 17.8％ (8/45).The positive predictive value of PI-RADS≥3 was 35.1％ (13/37) and 76.5％ (13/17) for prostate cancer by general radiologist and uroradiology specialist respectively,and for clinically significant prostate cancer,the positive predictive value of PI-RADS ≥ 3 was 21.6％ (8/37) and 52.9％ (9/17) respectively.Conclusions Uroradiology specialist achieved significantly superior in predictive value of PI-RADS for clinically significant prostate cancer compared with general radiologist.In the experienced centers,MRI-targeted biopsy could be performed only on high PI-RADS score lesions,thus to reduce unnecessary biopsies and to avoid over diagnosis and over treatment of prostate cancer.

Radiologist-technologist integration training program: Evaluation and practice / 医学研究生学报

The radiologist-technologist integration (RTI)training program involves radiologists′ participation in technologists′ scan and examination training and technologists′ participation in physicians′preparation of reports during their residency training.Based on the investigations in the PLA General Hospital, West China Hospital of Sichuan University, The SecondXiangyaHospitalofCentralSouthUniversity, International Hospital of Peking University, and The First Hospital of Qinhuangdao, the author evaluated the methods of and practice in RTItraining as well as the potential value of RTI and itsfacilitation of work quality.RTItraining has effectively enhanced the basic skills training of the residents, deepened their understanding of the imaging process, and promoted scientific research and clinical work.The RTI training program can cultivate the technologists′ abilities to identify lesions, design individual protocols, and achieve the targeted scan for diagnosis.Meanwhile, itmay contribute to the improvement of human resource management inthe radiology department and overall quality of radiological examinations.

Concordancia interobservador de hallazgos cardiopulmonares en la radiografía de tórax entre radiólogos y médicos generales de un servicio de urgencias / Interobserver agreement of cardiothoracic diseases in chest radiography between radiologists and general practitioners

Resumen Objetivo: estimar la concordancia inter-observador de hallazgos cardiopulmonares en la radiografía de tórax de adultos entre dos grupos independientes de radiólogos y médicos generales. Materiales y métodos: dos grupos de evaluadores, uno de radiólogos (n=2) y uno de médicos generales (n=5) valoraron 100 radiografías de tórax. Los ítems de evaluación fueron la calidad técnica radiológica, la normalidad de la radiografía y 27 hallazgos radiológicos comunes en la consulta de urgencias. Los evaluadores calificaron la posibilidad de encontrar los hallazgos específicos en la radiografía (cinco por cada placa) en un formato similar a un script de concordancia, con escala de respuesta tipo Likert. El cálculo de concordancia se realizó con el estadístico kappa por grupos de Vanbelle (K2g). Resultados: los grados de concordancia entre radiólogos y médicos generales fueron débiles para la identificación de hallazgos cardiopulmonares (K2g 0,46; IC 95 % 0,43 - 0,51), calidad de la imagen (K2g 0,44; IC 95 % 0,35 - 0,53) y determinación de normalidad (K2g 0,58; 0,44 - 0,72). Los índices de prevalencia fueron elevados (mín. - max.: 0,59 - 0,85) en la valoración de normalidad de las placas. Conclusiónes: El grado de acuerdo en la determinación de normalidad puede estar subestimado por un alto índice de prevalencia. El poco tiempo de formación en radiología y de experiencia en el campo de los médicos generales podrían estar asociados al bajo grado de acuerdo entre los grupos.

Abstract Objective: To estimate the inter-observer agreement between two independent groups of radiologists and general practitioners in the identification of cardiopulmonary findings via standard plain chest radiographs in adults. Materials and methods: Two groups of independent raters (Radiologists, n=2; General Practitioners, n=5) analyzed 100 chest radiographs according to the technical quality, normality, and 5 specific findings. Cardiopulmonary findings were registered via a script concordance-like test. We calculated agreement between groups with Vanbelle’s kappa coefficient (K2g). Results: The concordance between the groups of radiologists and general practitioners in specific chest x-ray findings (K2g 0.46, 95 %CI 0.43 - 0.51), image technical quality (K2g 0.44; 95 %CI 0.35 - 0.53), and normality (K2g 0.58; 95 %CI 0.44 - 0.72) was weak. Prevalence indices were high in the analysis of chest x-ray normality (min. - max.: 0.59 - 0.85). Conclusions: Kappa coefficients in the determination of normality could have been biased downward due to high prevalence indices. Short time of training in radiology and experience in the field could account for low agreement between the groups.

Radiologic Pattern of Sputum-positive Pulmonary Tuberculosis (PTB) among Immunocompetent Patients in Gwagwalada, Nigeria.

Background: The radiologic knowledge of tuberculosis-associated lung disease is an essential tool in the clinical diagnosis and management of tuberculosis, a disease that is still a big challenge to Nigeria and Africa at large. Chest radiogram is cheap and readily available, and usually the first imaging modality for the evaluation of patients with sputum positive pulmonary tuberculosis. Objective: To determine the radiologic pattern of sputum-positive PTB among immunocompetent patients in Gwagwalada, Nigeria. Methods: A cross-sectional study spanned sixteen months, involving one hundred adult patients recently clinically diagnosed of pulmonary tuberculosis were recruited and underwent chest radiographic examination with 14 × 17-inch or 17 × 17-inch image sizes for evaluation of pulmonary tuberculosis. Results: Of the 100 patients that underwent chest radiographs, ten patients (10.0%) had normal chest radiographs while varying degree of abnormal chest radiographic findings were seen in the remaining 90 (90%) patients. Fibrotic lesion were present in 45 (45.0%) of patients. There were 60 male and 40 female patients with eight males with normal chest radiogram. Fibrosis was the predominant radiological feature with 45 (45.0%) of all the patients examined in this study. The prevalence of other findings was: consolidation, 11 (11.0%); lung collapse, (11.0%); cavitation, (11.0%); pleural effusion, (7.0%) and lung infiltrate, (5.0%). Multiple lung cavities and military lesions were not observed in any patient. Cavities were present in 11 (11.0%) and all were solitary and located in the upper lung zone(s). Conclusion: Pulmonary tuberculosis is endemic in our environment and evaluation of radiology features in immunocompetent individual is encouraged in developing countries. Sequelae of tuberculosis such as fibrosis was frequently seen in this study which points to late presentation, therefore more public health advocacy are needed against pulmonary tuberculosis.

MRI Inter-Reader and Intra-Reader Reliabilities for Assessing Injury Morphology and Posterior Ligamentous Complex Integrity of the Spine According to the Thoracolumbar Injury Classification System and Severity Score

OBJECTIVE: To evaluate spine magnetic resonance imaging (MRI) inter-reader and intra-reader reliabilities using the thoracolumbar injury classification system and severity score (TLICS) and to analyze the effects of reader experience on reliability and the possible reasons for discordant interpretations. MATERIALS AND METHODS: Six radiologists (two senior, two junior radiologists, and two residents) independently scored 100 MRI examinations of thoracolumbar spine injuries to assess injury morphology and posterior ligamentous complex (PLC) integrity according to the TLICS. Inter-reader and intra-reader agreements were determined and analyzed according to the number of years of radiologist experience. RESULTS: Inter-reader agreement between the six readers was moderate (k = 0.538 for the first and 0.537 for the second review) for injury morphology and fair to moderate (k = 0.440 for the first and 0.389 for the second review) for PLC integrity. No significant difference in inter-reader agreement was observed according to the number of years of radiologist experience. Intra-reader agreements showed a wide range (k = 0.538-0.822 for injury morphology and 0.423-0.616 for PLC integrity). Agreement was achieved in 44 for the first and 45 for the second review about injury morphology, as well as in 41 for the first and 38 for the second review of PLC integrity. A positive correlation was detected between injury morphology score and PLC integrity. CONCLUSION: The reliability of MRI for assessing thoracolumbar spinal injuries according to the TLICS was moderate for injury morphology and fair to moderate for PLC integrity, which may not be influenced by radiologist' experience.

Venturas y desventuras del radiólogo de guardia / Fortunes and misfortunes of the radiologist on-call

El trabajo del radiólogo de guardia no es fácil, pero además el número de urgencias radiológicas se ha incrementado a lo largo de los años. Esto no sólo se debe a la gran cantidad de personas que acuden a los Servicios de Urgencias, sino también al aumento del espectro de exploraciones radiológicas para evaluar casos urgentes. Así, el gran volumen de pacientes como el elevado número de doctores, enfermeras y personal sanitario dando diversas instrucciones en la atención pueden transmitir, en ocasiones, a los Servicios de Emergencias la apariencia de un pseudo "campo de batalla". En este escenario, las vivencias del radiólogo de guardia, sus venturas y desventuras, resultan tan positivas como negativas. En este artículo, se presentan y discuten los factores y posibles situaciones problemáticas con mayor implicancia en el trabajo del radiólogo de guardia, así como también las potenciales medidas para abordarlas, evitarlas o resolverlas.

The job of the radiologist on-call is not easy and the number of radiological emergencies have increased over the years, but not only to the larger number of patients visiting the emergency departments, but also to broader spectrum of radiological's examinations requiered to asses emergency cases. The greater volume of patients, together with a larger number of doctors, nurses and other health professionals, giving diverse instructions for patients care, can at times give the emergency room the appearance of a pseudo-battlefield. The different experiences of the on-call radiologist in such setting can be both positive and negative. The contributing factors and possible measures for dealing with, avoiding or solving potentially problematic situations which, in turn, may create major implications for the on-call radiologist in carrying out his work are presented and discussed in this article.

Left ventricular ejection fraction measurement using cardiovascular magnetic resonance imaging in patients with post-myocardial infarction: assessment of reproducibility by a cardiovascular radiologist and a trained technologist.

Background: Cardiovascular magnetic resonance imaging (CMR) has recently been accepted as a preferential method for evaluation left ventricular ejection fraction (LVEF). The LVEF analysis by CMR is usually performed by trained technologists in many institutions of Thailand. Objective: Assess the reproducibility of LVEF measured by a cardiovascular radiologist and a trained technologist using CMR in patients with post-myocardial infarction (MI). Methods: Twenty-one MI patients (18 men and 3 women) were recruited, where nine patients underwent CMR and left ventriculography to follow-up LVEF two times in six months. Both CMR and left ventriculography were examined within two weeks. LVEF from CMR were measured by a cardiovascular radiologist and a trained technologist and the correlation between the left ventriculography and CMR was determined. Results: In 30 CMR studies, interobserver reliability (intraclass correlation coefficient ICC=0.94) and intraobserver reliability (ICC=0.96) was excellent. LVEF measured by left ventriculography was higher compared with that by CMR, and their correlation was moderate (ICC=0.56). Conclusion: The LVEF measurement by a cardiovascular radiologist and a trained technologist using CMR was very reproducible, but the correlation between CMR and left ventriculography was moderate.

Proposal on ROC-theory based quantitative assessment of radiologists'diagnostic competence / 中华医院管理杂志

Perfofinance of the radio-diagnostic system can be evaluated quantitatively by the analysis method of the ROC curve.Areas under the ROC curves represent the diagnostic perfogrmance efficiency.A comparison of the perforrmance variance of the diagnostic system incurred by different diagnosticians,will find the diagnostic competence of individual diagnosticians.This can help hospital management in deploying personnel in the department of radiology reasonably.

A Case of Chronic Radiodermatitis with CML and Squamous Cell Carcinoma in Radiologist

Chronic radiodermatitis may develop after repetitive exposure to small doses of ionizing radiation over an extensive period of time. Therefore, it is most likely to occur in people who work with ionizing radiation. The clinical symptoms of chronic radiodermatitis include atrophy, telangiectasia, sclerosis, pigmentary changes of the skin, ulceration and the development of malignancy. Cases of chronic radiodermatitis have occasionally been reported in physicians performing interventional procedures. Herein, we report a case of chronic radiodermatitis involving the 4th finger web and 5th finger of a radiologist, who suffered from chronic myelogenous leukemia shortly afterwards. The lesion on the 5th finger eventually progressed into squamous cell carcinoma.

Placement of Tunneled Cuffed Hemodialysis Catheters Via Internal Jugular Vein by an Interventional Radiologist and a Vascular Surgeon Together / 대한신장학회잡지

BACKGROUND: The use of dacron-cuffed tunneled double-lumen catheters for hemodialysis has become more common as patients wait for creation and maturation of a permanent access. Placement of the catheters is done by interventional radiologists, vascular surgeons or nephrologists, and the differences in success rates, complications, blood flow rates and the durations of catheter uses are reported. This study evaluated the usefulness, complications, blood flow rates and duration of use of tunneled cuffed hemodialysis catheters implanted via the internal jugular veins by an interventional radiologist and a vascular surgeon together. METHODS: The outcomes were retrospectively analyzed of 31 hemodialysis catheters placed from December 1999 through January 2001. We investigated age, sex, indications and locations of insertion, catheter performance, complications and causes of catheter removal. All the catheters were placed via the internal jugular veins by an interventional radiologist and a vascular surgeon together. RESULTS: Catheter placements were successful in all patients. Procedural complication was limited to clinically unimportant minor local bleeding. No instances of pneumothorax, hemothorax, vessel injury, substantial bleeding, obstruction, malposition or stenosis occurred. The blood flow rate on the first hemodialysis after placement of the catheter was 230+/-35.5 mL/min and that after 1 month was 248+/-18.6 mL/min. Late complications included catheter breakage in two cases (6.5%) and bacteremia in four cases (12.9%). Catheters were removed because of catheter-related bacteremia in four cases (12.9%) and death of patients in three cases (9.7%). In 22 cases (71.0%) the catheters were removed because the A- V vascular accesses were available for hemodialysis. Mean duration of the catheter use before removal was 134+/-96 days. CONCLUSION: Tunneled cuffed catheters inserted via the internal jugular veins are safe and durable vascular accesses for hemodialysis with good blood flow rate and long duration of use, especially placed by the co-operation of an interventional radiologist and a vascular surgeon.

Future Supply of and Demand for Diagnostic Radiologists in Korea

PURPOSE: To project the supply and demand of diagnostic radiologists in Korea. MATERIALS AND METHODS: With the year 2010 in mind, the 'in-and-out moves' method was applied in order to estimate the supply of diagnostic radiologists. We also used the health need approach and the physician population ratio method for the projection of demand. As sources of information, administrative and demographic statistics were analyzed. RESULTS: If the current situation in terms of the supply of newly qualified diagnostic radiologists is maintained, the projected number will be in short supply by as many as 790 in 2000 and by 370 in 2005. There will, however, be an oversupply of about 110 in 2010. CONCLUSION: To accomplish a balanced state of supply and demand in 2010, the numbers of newly qualifying diagnostic radiologists should be maintained at the current level.

Impact of clinical history on film interpretation

We performed a study to determine whether clinical history gives a positive or negative influence on X-ray film interpretation. One hundred and nine patient's radiograms, consisting of 55 normal and 54 abnormal cases (136 abnormalities), were interpreted twice by three pairs of residents in radiology and a pair of qualified radiologists, without clinical history first and with clinical history next. The interpreters recorded diagnosis and confidence level of normal or abnormal findings on a six-point scale. Analysis of receiver operating characteristic (ROC) curves showed that knowledge of clinical history improved diagnostic accuracy. Residents, especially beginners, should be advised to obtain clinical history whenever they read radiograms.

AN OBSERVATION ON MICRONUCLEAR CELL RATE IN NUCLEATED ERYTHROCYTE OF BONE MARROW OF RADIOLOGISTS / 重庆医科大学学报

The micronuclear cell rate of nucleated erythrocyte of bone marrow in 30 cases of radiologists and 20 cases of normal controls have been compared. The result showed that the mioronuclear cell rate in bone marrow of radiologists is higher than that of normal contrls and the difference is significant(P0.06).