A Review of Radiology Reports From Hip Surveillance X-Rays Completed in Community Hospitals

Deep learning models have received much attention lately for their ability to achieve expert-level performance on the accurate automated analysis of chest X-rays. Although publicly available chest X-ray datasets include high resolution images, most models are trained on reduced size images due to limitations on GPU memory and training time. As compute capability continues to advance, it will become feasible to train large convolutional neural networks on high-resolution images. This study is based on the publicly available MIMIC-CXR-JPG dataset, comprising 377,110 high resolution chest X-ray images, and provided with 14 labels to the corresponding free-text radiology reports. We find, interestingly, that tasks that require a large receptive field are better suited to downscaled input images, and we verify this qualitatively by inspecting effective receptive fields and class activation maps of trained models. Finally, we show that stacking an ensemble across resolutions outperforms each individual learner at all input resolutions while providing interpretable scale weights, suggesting that multi-scale features are crucially important to information extraction from high-resolution chest X-rays.

Download Full-text

Worsening bone scan in the evaluation of antitumor response during hormonal therapy of breast cancer.

Journal of Clinical Oncology ◽

10.1200/jco.1995.13.5.1123 ◽

1995 ◽

Vol 13 (5) ◽

pp. 1123-1128 ◽

Cited By ~ 82

Author(s):

C L Vogel ◽

J Schoenfelder ◽

I Shemano ◽

D F Hayes ◽

R A Gams

Keyword(s):

Breast Cancer ◽

Disease Progression ◽

Retrospective Analysis ◽

Hormonal Therapy ◽

Bone Scan ◽

Large Scale ◽

Response To Therapy ◽

X Rays ◽

Cancer Trial ◽

Radiology Reports

PURPOSE Scintigraphic flare in association with response to therapy has been well described in the medical literature. During the course of a recent breast cancer trial, it became apparent that several patients with worsening bone scan but no other clinical evidence of disease progression might have potentially benefited from continued therapy, but had therapy discontinued. A retrospective analysis of this issue was performed to assess the magnitude and scope of this problem. MATERIALS AND METHODS A total of 648 patients with hormone receptor-positive or unknown advanced breast cancer were treated as part of a large-scale trial of first-line hormonal therapy. Patients were assessed for response to therapy, including response duration, progression-free interval (PFI), overall survival, and quality of life. The retrospective analysis presented here was performed to assess whether patients with a possible scintigraphic flare within the first 16 weeks of therapy might have had therapy discontinued prematurely due to a worsening bone scan attributable to tumor flare, rather than due to disease progression. RESULTS Analysis of the hormonal trial showed that of 376 assessable patients 108 (29%) with bone disease had a possible scintigraphic flare by week 8 or 16 of the trial, based on data on the case report forms and radiology reports (bone scans and x-rays). Of these, 69 patients (64%) were continued on study therapy, which resulted in clinical benefit in 50 (72%) of those patients. In contrast, 39 patients (36%) with possible scintigraphic flare were removed from the trial. CONCLUSION We conclude that changes in bone scintigraphy that mimic progressive disease early in the course of hormonal treatment of patients with breast cancer metastatic to bone may represent scintigraphic flare associated with response. Thus, clinicians must be cognizant of the phenomenon of scintigraphic flare to avoid premature discontinuation of a potentially beneficial treatment.

Download Full-text

Development of a System that Generates Structured Reports for Chest X-ray Radiography

Methods of Information in Medicine ◽

10.3414/me09-01-0014 ◽

2010 ◽

Vol 49 (04) ◽

pp. 360-370 ◽

Cited By ~ 4

Author(s):

Y. Matsumura ◽

N. Mihara ◽

Y. Kawakami ◽

K. Sasai ◽

H. Takeda ◽

...

Keyword(s):

Narrative Form ◽

Free Text ◽

X Rays ◽

Sentence Structure ◽

X Ray ◽

Radiology Reports ◽

Chest X Ray ◽

Structured Reports ◽

Description Framework ◽

Resource Description

Summary Objectives: Radiology reports are typically made in narrative form; this is a barrier to the implementation of advanced applications for data analysis or a decision support. We developed a system that generates structured reports for chest x-ray radiography. Methods: Based on analyzing existing reports, we determined the fundamental sentence structure of findings as compositions of procedure, region, finding, and diagnosis. We categorized the observation objects into lung, mediastinum, bone, soft tissue, and pleura and chest wall. The terms of region, finding, and diagnosis were associated with each other. We expressed the terms and the relations between the terms using a resource description framework (RDF) and developed a reporting system based on it. The system shows a list of terms in each category, and modifiers can be entered using templates that are linked to each term. This system guides users to select terms by highlighting associated terms. Fifty chest x-rays with abnormal findings were interpreted by five radiologists and reports were made either by the system or by the free-text method. Results: The system decreased the time needed to make a report by 12.5% compared with the free-text method, and the sentences generated by the system were well concordant with those made by free-text method (F-measure = 90%). The results of the questionnaire showed that our system is applicable to radiology reports of chest x-rays in daily clinical practice. Conclusions: The method of generating structured reports for chest x-rays was feasible, because it generated almost concordant reports in shorter time compared with the free-text method.

Download Full-text

UTILIZATION OF RADIOLOGY REPORTS AT COMMUNITY HOSPITALS IN GUNMA

The KITAKANTO Medical Journal ◽

10.2974/kmj1951.43.657 ◽

1993 ◽

Vol 43 (6) ◽

pp. 657-665

Author(s):

KUNIAKI TOMIOKA ◽

HIDEKI SUZUKI ◽

TOMIO INOUE ◽

MITSUOMI MATSUMOTO ◽

KEIGO ENDO

Keyword(s):

Community Hospitals ◽

Radiology Reports

Download Full-text

278 The Acute Spine Fracture in the Emergency Department; a Retrospective Review

Age and Ageing ◽

10.1093/ageing/afz103.174 ◽

2019 ◽

Vol 48 (Supplement_3) ◽

pp. iii17-iii65

Author(s):

Owen Thorpe ◽

Avril Beirne ◽

Paul Fox ◽

Aoife Nic Uidhir ◽

Frances Dockery

Keyword(s):

Best Practice ◽

Osteoporosis Treatment ◽

High Risk Group ◽

Consecutive Series ◽

Spine Fractures ◽

Acute Management ◽

X Rays ◽

Available N ◽

Radiology Reports ◽

Fracture Diagnosis

Abstract Background As to who should manage osteoporotic spine fractures presenting to Emergency departments (ED) is sometimes debated. We sought to review practice regarding their management in our institution that might inform a clinical pathway. Methods We conducted a search of radiology reports for a consecutive series of thoracic or lumbar spine x-rays ordered by ED team only, i.e. a series of patients whose presenting complaint was a suspected acute fracture. We narrowed search to include terms ‘compression’ or ‘wedge’ or ‘end-plate’. Results Over 7 months, there were 1,505 such reports; narrowed search and excluding duplicates yielded 168 patients of whom 84 had a fracture. We looked at the acute management of those >50yrs, excluding one metastatic fracture leaving n=64. Of these, 65% occurred following a fall, 14% on twisting/bending/coughing, 14% spontaneous, 7% unclear onset. ED first consulted orthopaedics for 10 cases, neurosurgeons for 2, physicians for 18. A total of 21 were discharged from ED (5 having speciality review pre-discharge). A further 11 were sent home from ED with fracture diagnosis made only when x-ray subsequently reported. Of those admitted, 28 went to physicians with consult to surgeons in 39%, 2 went to orthopaedic surgery, 2 to other specialists. Admission was complicated by pressure ulcer in 13% (4/32), pneumonia in 13%. Overall 24/64 (38%) went on to have MRI/CT (mainly admitted cases). At least 12 were managed with a brace (all records not available). N=7 (11%) had later vertebroplasty. More than half had no documented osteoporosis treatment plans nor GP instruction to address. Conclusion Care of spine fractures presenting acutely varies; a high proportion managed by ED solely. Whether outcomes vary as a result is not answered by this audit but there is a need for a pathway to inform best practice. Osteoporosis is inadequately-addressed in this high risk group, highlighting need for fracture liaison services in post-acute management.

Download Full-text

X-ray checks of NG tube position: a case for guided tube placement

British Journal of Radiology ◽

10.1259/bjr.20210432 ◽

2021 ◽

pp. 20210432

Author(s):

Stephen Taylor ◽

Alex R Manara

Keyword(s):

Time Delay ◽

Drug Treatment ◽

Tube Placement ◽

X Rays ◽

X Ray ◽

Factors Associated ◽

Radiology Reports ◽

Tube Position ◽

Clinically Significant ◽

Combined Methods

Objectives: Checking nasogastric (NG) tube position by X-ray is too late to prevent 1.5% of blind tube placements entering the lung and results in delays to feeding and drugs. We audit the safety of the tube position and delay incurred by X-ray. Methods: From Radiology reports, we determined whether tube position was safe for feeding, factors associated with an X-ray request and the time delay from X-ray request to that report. For tubes misplaced into the lung, the distance from the carina to tube tip was measured and compared with that from published records of guided tube placement. Results: From 1 July 2019 to 30 June 2020, 1934 X-rays were done to check NG tube position in 891 patients. Gastric placement was confirmed in 85% but, because of tube proximity to the oesophagus, only 73% were deemed safe to feed. The 2.2% of tubes reported to be in the lung were a median of 18 cm beyond the carina compared to 12 cm and 0 cm for electromagnetic and direct vision methods of guided placement. X-ray checks delayed feed and drug treatment by >2 h in 51% of placements and 33% of patients required >3 X-rays during their enteral episode. Conclusion: X-ray checks are common and detect a high percentage of unsafe tube placements, leading to repeated X-ray and delayed delivery of drugs and nutrition. Interpretation can be difficult even when following standard national criteria and post-placement X-ray cannot prevent deep lung placement. Guided or combined methods of confirming tube placement should be investigated. Advances in knowledge: Reports included 27.5% of placements as unsafe, 2.2% in the lung at a median depth of 18 cm beyond the carina and too late to prevent 7 pneumothoraces. X-rays were repeated >3 times in 33% of patients over their enteral course and we are associated with clinically significant delays to drug treatment (and nutrition) in 51%; combined methods of tube confirmation or guided placement may be safer and more efficient.

Download Full-text

Assessing spontaneous passage of prophylactic pancreatic duct stents by X-ray: is a radiology report adequate?

Therapeutic Advances in Gastrointestinal Endoscopy ◽

10.1177/2631774519862895 ◽

2019 ◽

Vol 12 ◽

pp. 263177451986289

Author(s):

Justin Loloi ◽

Jacob S. Lipkin ◽

Eileen M. Gagliardi ◽

John M. Levenick

Keyword(s):

Pancreatic Duct ◽

False Negative ◽

Retrospective Chart Review ◽

Pancreatic Ducts ◽

X Rays ◽

X Ray ◽

Radiology Reports ◽

Spontaneous Passage ◽

Pancreatic Duct Stent

Background: Pancreatic duct stents are frequently placed for prophylaxis of post-endoscopic retrograde cholangiopancreatography pancreatitis. Because of concern for possible secondary ductal changes from a retained stent, these stents need to be monitored and removed if retained. Usually an abdominal X-ray is performed to assess retained stent, and if present, an esophagogastroduodenoscopy is performed to remove the stent. Limited data is published on false-negative radiology reports for spontaneous passage of stents. Methods: Using an Institutional Review Board–approved stent log, a retrospective chart review of all pancreatic duct stents placed at our institution from 2008 to 2014 was performed. Results: A total of 856 pancreatic duct stents were placed during the study period. Of these, 435 (50.8%) were prophylactic stents and 421 (49.2%) were therapeutic. Complete follow-up data were available in 426 (97.9%) patients with prophylactic stents. Six patients (1.4%) were lost to follow up and three (0.7%) expired prior to removal. In all, 283 (66%) had follow-up imaging, with 167 (39.2%) having the official radiology read with no retained pancreatic duct stent in place. Eight of these cases were “false-negative” radiology interpretation (4.8% of cases read as “no stent,” NNH = 20). The stent was found either by review of image by an endoscopist or incidental stent discovery during a follow-up procedure. Conclusion: Radiologist interpretation of abdominal X-rays to assess spontaneous passage of prophylactic pancreatic ducts stents resulted in a false-negative interpretation in approximately 5% of cases. Independent review of the images by the endoscopist may be beneficial given unfamiliarity of these stents by radiologists.

Download Full-text

Labelling chest x-ray reports using an open-source NLP and ML tool for text data binary classification

10.1101/19012518 ◽

2019 ◽

Author(s):

Sohrab Towfighi ◽

Arnav Agarwal ◽

Denise Y. F. Mak ◽

Amol Verma

Keyword(s):

Internal Medicine ◽

Language Processing ◽

Performance Metrics ◽

Random Search ◽

General Medicine ◽

X Rays ◽

Text Data ◽

X Ray ◽

Radiology Reports ◽

Chest X Ray

AbstractThe chest x-ray is a commonly requested diagnostic test on internal medicine wards which can diagnose many acute pathologies needing intervention. We developed a natural language processing (NLP) and machine learning (ML) model to identify the presence of opacities or endotracheal intubation on chest x-rays using only the radiology report. This a preliminary report of our work and findings. Using the General Medicine Inpatient Initiative (GEMINI) dataset, housing inpatient clinical and administrative data from 7 major hospitals, we retrieved 1000 plain film radiology reports which were classified according to 4 labels by an internal medicine resident. NLP/ML models were developed to identify the following on the radiograph reports: the report is that of a chest x-ray, there is definite absence of an opacity, there is definite presence of an opacity, the report is a follow-up report with minimal details in its text, and there is an endotracheal tube in place. Our NLP/ML model development methodology included a random search of either TF-IDF or bag-of-words for vectorization along with random search of various ML models. Our Python programming scripts were made publicly available on GitHub to allow other parties to train models using their own text data. 100 randomly generated ML pipelines were compared using 10-fold cross validation on 75% of the data, while 25% of the data was left out for generalizability testing. With respect to the question of whether a chest x-ray definitely lacks an opacity, the model’s performance metrics were accuracy of 0.84, precision of 0.94, recall of 0.81, and receiver operating characteristic area under curve of 0.86. Model performance was worse when trained against a highly imbalanced dataset despite the use of an advanced oversampling technique.

Download Full-text

An artificial intelligence deep learning model for identification of small bowel obstruction on plain abdominal radiographs

British Journal of Radiology ◽

10.1259/bjr.20201407 ◽

2021 ◽

pp. 20201407

Author(s):

DH Kim ◽

H Wit ◽

M Thurston ◽

M Long ◽

GF Maskell ◽

...

Keyword(s):

Deep Learning ◽

Small Bowel ◽

Small Bowel Obstruction ◽

Bowel Obstruction ◽

Learning Model ◽

X Rays ◽

Automated Identification ◽

Radiology Reports ◽

Abdominal Radiographs ◽

Deep Learning Model

Objectives: Small bowel obstruction is a common surgical emergency which can lead to bowel necrosis, perforation and death. Plain abdominal X-rays are frequently used as a first-line test but the availability of immediate expert radiological review is variable. The aim was to investigate the feasibility of using a deep learning model for automated identification of small bowel obstruction. Methods: A total of 990 plain abdominal radiographs were collected, 445 with normal findings and 445 demonstrating small bowel obstruction. The images were labelled using the radiology reports, subsequent CT scans, surgical operation notes and enhanced radiological review. The data were used to develop a predictive model comprising an ensemble of five convolutional neural networks trained using transfer learning. Results: The performance of the model was excellent with an area under the receiver operator curve (AUC) of 0.961, corresponding to sensitivity and specificity of 91 and 93% respectively. Conclusion: Deep learning can be used to identify small bowel obstruction on plain radiographs with a high degree of accuracy. A system such as this could be used to alert clinicians to the presence of urgent findings with the potential for expedited clinical review and improved patient outcomes. Advances in knowledge: This paper describes a novel labelling method using composite clinical follow-up and demonstrates that ensemble models can be used effectively in medical imaging tasks. It also provides evidence that deep learning methods can be used to identify small bowel obstruction with high accuracy.

Download Full-text