Computed Tomography Reports Research Articles

Clinically relevant findings on 24-h head CT after acute stroke therapy: The 24-h CT score.

Routine head computed tomography (CT) is performed 24 h post-acute stroke thrombolysis and thrombectomy, even in patients with stable or improving clinical deficits. Predicting CT results that impact management could help prioritize patients at risk and potentially reduce unnecessary imaging. In this institutional review board (IRB)-approved retrospective study, data from 1461 consecutive acute ischemic stroke patients at our Comprehensive Stroke Center (n = 8943, 2012-2022) who received intravenous thrombolysis or endovascular therapy, exhibited stable or improving 24-h exams, and underwent 24-h follow-up head CT per standard acute stroke care guidelines. CT reports 24 h post-stroke were reviewed for edema, mass effect, herniation, and hemorrhage. The primary outcome was any clinically relevant 24-h CT finding that led to changes in antithrombotic treatment or blood pressure goals, extended intensive care unit (ICU) stays or hospitalizations, neurosurgical interventions, or administration of mannitol or hypertonic saline. Multivariable logistic regression identified independent predictors of clinically meaningful CT abnormalities. A 24-h CT score was developed and cross-validated. The mean age was 70 years, with 47% women. The median National Institutes of Health Stroke Scale (NIHSS) score at admission was 12 (interquartile range (IQR): 6-18). Stroke-related abnormalities on 24-h CT were present in 325 patients (22.2%), with 183 (12.5%) showing clinically relevant findings. Age, admission NIHSS, and blood glucose levels were independent predictors of clinically relevant 24-h CT findings. The final model C statistic was 0.72 (95% confidence interval (CI): 0.68-0.76) in the derivation cohort and 0.72 (95% CI: 0.67-0.75) in bootstrapping validation. The 24-h CT score was developed using these predictors: NIHSS score 5-15 (+3); NIHSS score ⩾16 (+5); age < 75 years (+1); admission glucose ⩾ 140 mg/dL (+1). The prevalence of clinically relevant CT findings was 4.3% in the low-risk group (24-h CT score ⩽ 4), 11.3% in the medium-risk group (score 5), and 21.4% in the high-risk group (score ⩾ 6). The 24-h CT score demonstrated good calibration. In patients undergoing thrombolysis or thrombectomy who undergo routine 24-h head CT despite remaining clinically stable or improving, only one in eight prove to have 24-h head CT findings that impact management. The 24-h CT score provides risk stratification that may improve resource utilization. A.S. and B.Z. have full access to the data used in the analysis in this article. Deidentified data will be shared after ethics approval if requested by other investigators for purposes of replicating the results.

Pleural and pericardial effusions as prognostic factors in patients with acute pulmonary embolism: a multicenter study.

The prognostic role of pleural and pericardial effusion in patients with acute pulmonary embolism (PE) is still unclear with a trend for worse clinical outcome. The aim of the present study was to demonstrate the prognostic role of pleural and pericardial effusion in patients with acute PE in a large multicentre setting. The investigated patient sampled was retrospectively comprised of 1082 patients (494 female, 45.7%) with a mean age of 63.8 years ± 15.8. In every case, contrast enhanced computed tomography (CT) pulmonalis angiography was analyzed to diagnose and quantify the pleural and pericardial effusion. The 30-day mortality was the primary endpoint of this study. A total of 127 patients (11.7%) died within the 30-day observation period. Pleural effusion was identified in 438 patients (40.5%) and pericardial effusion was identified in 196 patients (18.1%). The presence of pleural effusion was associated with 30-day mortality, HR = 2.78 (95%CI1.89-4.0), p < 0.001 (univariable analysis), and HR = 2.52 (95%CI1.69-3.76), p < 0.001 (multivariable analysis). The pleural effusion width and density were not associated with 30-day mortality. The presence of pericardial effusion was not associated with 30-day mortality in multivariable analysis, HR = 1.28 (95%CI 0.80-2.03), p = 0.29. Pleural effusion is a common finding in patients with acute pulmonary embolism, occurring in 40.5% of cases, and is a prognostic imaging finding associated with 30-day mortality. The presence of pleural effusion alone, regardless of volume or density, has been shown to be prognostic and should be included in CT reports. The prognostic role of pericardial effusion is limited.

Comparison of the Effectiveness of Computed Tomography and Magnetic Resonance Imaging Techniques in Patient Groups Aged under and over 65 Years and Diagnosed with Ischemic Stroke in the Emergency Department

Objectives: Stroke is a condition with high morbidity and mortality. This study aims to investigate whether the effectiveness of Computed Tomography (CT) and diffusion-weighted Magnetic Resonance Imaging (MRI), the techniques that have a significant place in the diagnosis of ischemic stroke, the most common form of stroke, are affected by the physiological changes of advanced age. Methods: A total of 436 patients were included in the study. The study population was divided into two groups depending on age: those above 65 and those under 65. Medical files, both the emergency department and clinical ward files, of the patients who were admitted to the emergency department in nine months and admitted to the neurology clinic with the diagnosis of ischemic stroke were retrospectively analyzed. The time from admission to imaging was determined depending on patient files and the Hospital Management Information System (HBYS). The CT and MRI reports interpreted by radiologists were also reviewed. While recording the data, the presence of a lesion, its direction, and localization were also noted. Results: CT positivity was 21.3%, and the positivity of diffusion-weighted imaging was 82.1% in the study population. The time was shorter in the group of patients with positive CT results than in the group with negative CT results. In subjects under 65, the time between onset and imaging was shorter in the group with positive CT results than in the group with negative CT results. In subjects over 65, the time with positive CT results was not different from the group with negative CT results. It was determined that the mean time was shorter in the group with positive MRI results than in the group with negative results. In both the groups under the age of 65 and over the age of 65, the time interval was shorter in the patients with positive MRI results compared to those with negative MRI results. Conclusions: Regardless of the positivity or negativity of CT and MRI results, the mean time from symptom onset to imaging was shorter in the group under 65 years of age compared to the group over 65 years. Aging prolongs the time to admission and the neuroradiological response of geriatric patients.

Decoding Radiology Reports: Artificial Intelligence-Large Language Models Can Improve the Readability of Hand and Wrist Orthopedic Radiology Reports.

The purpose of this study was to assess the effectiveness of an Artificial Intelligence-Large Language Model (AI-LLM) at improving the readability of hand and wrist radiology reports. The radiology reports of 100 hand and/or wrist radiographs, 100 hand and/or wrist computed tomography (CT) scans, and 100 hand and/or wrist magnetic resonance imaging (MRI) scans were extracted. The following prompt command was inserted into the AI-LLM: "Explain this radiology report to a patient in layman's terms in the second person: [Report Text]." The report length, Flesch reading ease score (FRES), and Flesch-Kincaid reading level (FKRL) were calculated for the original radiology report and the AI-LLM-generated report. The accuracy of the AI-LLM report was assessed via a 5-point Likert scale. Any "hallucination" produced by the AI-LLM-generated report was recorded. There was a statistically significant improvement in mean FRES scores and FKRL scores in the AI-LLM-generated radiograph report, CT report, and MRI report. For all AI-LLM-generated reports, the mean reading level improved to below an eighth-grade reading level. The mean Likert score for the AI-LLM-generated radiograph report, CT report, and MRI report was 4.1 ± 0.6, 3.9 ± 0.6, and 3.9 ± 0.7, respectively. The hallucination rate in the AI-LLM-generated radiograph report, CT report, and MRI report was 3%, 6%, and 6%, respectively. This study demonstrates that AI-LLM effectively improves the readability of hand and wrist radiology reports, underscoring the potential application of AI-LLM as a promising and innovative patient-centric strategy to improve patient comprehension of their imaging reports.Level of Evidence: IV.

Revolutionizing Intracranial Hemorrhage Diagnosis: A Retrospective Analytical Study of Viz.ai ICH for Enhanced Diagnostic Accuracy.

Introduction Artificial intelligence (AI) alerts the radiologist to the presence of intracranial hemorrhage (ICH) as fast as 1-2 minutes from scan completion, leading to faster diagnosis and treatment. We wanted to validate a new AI application called Viz.ai ICH to improve the diagnosis of suspected ICH. Methods We performed a retrospective analysis of 4,203 consecutive non-contrast brain computed tomography (CT) reports in a single institution between September 1, 2021, and January 31, 2022. The reports were made by neuroradiologists who reviewed each case for the presence of ICH. Reports and identified cases with positive findings for ICH were reviewed. Positive cases were categorized based on subtype, timing, and size/volume. Viz.ai ICH output was reviewed for positive cases. This AI model was validated by assessing its performance with Viz.ai ICH as the index test compared to the neuroradiologists' interpretation as the gold standard. Results According to neuroradiologists, 9.2% of non-contrast brain CT reports were positive for ICH. The sensitivity of Viz.ai ICH was 85%, specificity was 98%, positive predictive value was 81%, and negative predictive value was 99%. Subgroup analysis was performed based on intraparenchymal, subarachnoid, subdural, and intraventricular subtypes. Sensitivities were 94%, 79%, 83%, and 44%, respectively. Further stratification revealed sensitivity improves with higher acuity and volume/size across subtypes. Conclusion Our analysis indicates that AI can accurately detect ICH's presence, particularly for large-volume/large-size ICH. The paper introduces a novel AI model for detecting ICH. This advancement contributes to the field by revolutionizing ICH detection and improving patient outcomes.

The prevalence of incidental findings in computed tomography of the head in Pediatric Emergency Department

Introduction and aim. Computed tomography (CT) is the first imaging modality in the evaluation of children in case of patients with head injury in pediatric Emergency Departments (EDs). Radiological CT reports include not only lesions that are the main cause of the child’s complaints but also incidental findings. The objective of this study is to assess incidental findings observed in children who were admitted to the ED and had the head CT performed. Material and methods. This retrospective, cross-sectional study enrolled 644 children under the age of 18, from 1st January 2021 to 31st June 2021. Each child could have had one or more incidental findings in CT Results. Among all CT studies, incidental findings were found in 279 out of 644 (43.32%) cases, of which 73 (11.34%) had both lesions detected incidentally and related to the trauma. Conclusion. Head CT is an incredibly useful tool in the assessment of some head emergencies. However, evaluation of the prevalence of incidental findings is difficult. Most of them require no specific further investigation. Pediatricians, who order CTs in children, must be prepared to interpret and communicate findings to families and introduce treatment in necessary situations.

Does timely reporting of preoperative CT scans influence outcomes for patients following emergency laparotomy?

Timely preoperative computed tomography (CT) scans are important for patients requiring emergency laparotomy. United Kingdom guidelines state that a CT scan should be reported within 1h for 'critical' patients (will alter management at the time) and within 12h for 'urgent' patients (will alter management but not necessarily that day). An observational study included patients who were added to the National Emergency Laparotomy Audit (NELA) at a National Health Service trust from 2014 to 2021. The association of compliance with timings guidance and mortality was investigated. Multivariable logistic regression was used to determine the odds ratio of adherence to guidelines according to age, gender, night time admission, American Society of Anesthesiology (ASA) score, NELA mortality risk and category of scan. Further models determined the influence of adherence to guidelines on mortality, also adjusted for these variables. There were 1,299 patients (48% 'critical' and 52% 'urgent' CT scans). Only 360/1,299 (28%) of scans were undertaken with adherence to the timing guidelines. Critical scans were less likely to adhere to guidelines. Although univariable analysis suggested that adherence to guidelines was associated with reduced mortality, this was not the case in the multivariable model: only age, ASA and NELA mortality risk remained significantly associated with mortality. A minority of patients met the recommended preoperative CT report timings, and this was less likely for scans designated 'critical'. This did not appear to affect mortality when adjusted for key variables of risk. This illustrates the phenomenon of guideline adherence appearing to affect patient outcomes as a product of selection bias rather than causality.

Are renal stone protocol computed tomography reports giving us enough information?

Non-contrast computed tomography (CT) is the gold-standard diagnostic test for urolithiasis. Little is published regarding which information needs to be included in the report for it to be most useful to the healthcare team for efficient triage and high-quality patient care. This study aimed to assess the quality and variability of CT scan reporting at a single Canadian tertiary academic medical center. We completed a retrospective review of 100 consecutive renal colic CT scans. Descriptive statistics were used to report the frequency with which specific elements commonly used by urologists to triage and treat patients were included in radiology reports. Our sample had a mean age of 51.4±13.1 years. Stone size was universally reported for obstructing stones but was less frequently reported for non-obstructing stones (100% vs. 86.8%). A similar trend was observed for the exact stone number (100% vs. 93.4%). Non-obstructing stones were more likely than obstructing stones to be reported in one dimension (77.5% vs. 47%). Obstructing stones were reported in three dimensions 27% of the time. CT reports commonly include the presence or absence of hydronephrosis status (98%) but are less likely to include renal size (32%) and periureteral stranding (16%). Hounsfield units (HU) were reported in 3% of the reports, but skin-to-stone distance (SSD) and radiation dose were never reported. Reports routinely included assessments of stone size, location, and number (although not uniformly). HU, SSD, and radiation dose were rarely reported. This provides insight into opportunities for standardized reporting to optimize knowledge transfer that may result in clinical efficiency and improved quality of patient care.

Natural language processing for automated breast cancer recurrence detection and classification in computed tomography reports.

e13591 Background: Relapse is a major concern for oncologists and breast cancer survivors that necessitates additional treatment and often leads to mortality. Cancer registries routinely track cancer mortality, but few monitor for relapse because of logistical challenges and prohibitive costs. In this context, Natural Language Processing (NLP) is a promising tool. Merging artificial intelligence with linguistics, NLP can rapidly analyze vast volumes of text in electronic health records. This capability of NLP is particularly valuable for Computed Tomography (CT) scans used in breast cancer care. CT scans are routinely used to characterize breast cancer progression and are described in transcribed dictations by radiologists. We aimed to apply NLP to these text reports to identify breast cancer relapses. Objective: To automate breast cancer relapse detection and classification in CT text reports using NLP. Methods: We analyzed 1,445 CT text reports from patients diagnosed with breast cancer between January 1, 2005, and December 31, 2014. These reports underwent manual review by trained human annotators. Text was annotated to identify terminology defining local, regional, and distant breast cancer relapses. Annotated reports were partitioned into a training-validation set (90% cohort) and a test set (10% cohort) for NLP model development. Results: In our dataset of 1,445 CT text reports, 72 (5.0%) were classified as local relapse, 97 (6.7%) as regional relapse, and 743 (51.4%) as distant relapse. The performance of our NLP model using the training-validation dataset can be summarized by the following metrics and 95% confidence intervals: 94% (±3.2) accuracy for detection and 96% (±2.9) accuracy for classification. The performance of our NLP model was confirmed using the test dataset, with 90% (±4.5) accuracy for detection and 91% (±6.3) accuracy for classification. For reference, all metrics are outlined (Table). Conclusions: Our model for identifying regional and distant relapses in CT reports had excellent performance, but had lower sensitivity for local relapses, posing a risk of false negatives. Automating the identification and classification of breast cancer relapses, if used retrospectively, can enhance cancer registry data about patient outcomes and, if used prospectively, holds the potential for enhancing patient care. [Table: see text]

Use of Natural Language Processing to Infer Sites of Metastatic Disease From Radiology Reports at Scale.

To evaluate natural language processing (NLP) methods to infer metastatic sites from radiology reports. A set of 4,522 computed tomography (CT) reports of 550 patients with 14 types of cancer was used to fine-tune four clinical large language models (LLMs) for multilabel classification of metastatic sites. We also developed an NLP information extraction (IE) system (on the basis of named entity recognition, assertion status detection, and relation extraction) for comparison. Model performances were measured by F1 scores on test and three external validation sets. The best model was used to facilitate analysis of metastatic frequencies in a cohort study of 6,555 patients with 53,838 CT reports. The RadBERT, BioBERT, GatorTron-base, and GatorTron-medium LLMs achieved F1 scores of 0.84, 0.87, 0.89, and 0.91, respectively, on the test set. The IE system performed best, achieving an F1 score of 0.93. F1 scores of the IE system by individual cancer type ranged from 0.89 to 0.96. The IE system attained F1 scores of 0.89, 0.83, and 0.81, respectively, on external validation sets including additional cancer types, positron emission tomography-CT ,and magnetic resonance imaging scans, respectively. In our cohort study, we found that for colorectal cancer, liver-only metastases were higher in de novo stage IV versus recurrent patients (29.7% v 12.2%; P < .001). Conversely, lung-only metastases were more frequent in recurrent versus de novo stage IV patients (17.2% v 7.3%; P < .001). We developed an IE system that accurately infers metastatic sites in multiple primary cancers from radiology reports. It has explainable methods and performs better than some clinical LLMs. The inferred metastatic phenotypes could enhance cancer research databases and clinical trial matching, and identify potential patients for oligometastatic interventions.

Comparison of natural language processing algorithms in assessing the importance of head computed tomography reports written in Japanese

PurposeTo propose a five-point scale for radiology report importance called Report Importance Category (RIC) and to compare the performance of natural language processing (NLP) algorithms in assessing RIC using head computed tomography (CT) reports written in Japanese.Materials and methods3728 Japanese head CT reports performed at Osaka University Hospital in 2020 were included. RIC (category 0: no findings, category 1: minor findings, category 2: routine follow-up, category 3: careful follow-up, and category 4: examination or therapy) was established based not only on patient severity but also on the novelty of the information. The manual assessment of RIC for the reports was performed under the consensus of two out of four neuroradiologists. The performance of four NLP models for classifying RIC was compared using fivefold cross-validation: logistic regression, bidirectional long–short-term memory (BiLSTM), general bidirectional encoder representations of transformers (general BERT), and domain-specific BERT (BERT for medical domain).ResultsThe proportion of each RIC in the whole data set was 15.0%, 26.7%, 44.2%, 7.7%, and 6.4%, respectively. Domain-specific BERT showed the highest accuracy (0.8434 ± 0.0063) in assessing RIC and significantly higher AUC in categories 1 (0.9813 ± 0.0011), 2 (0.9492 ± 0.0045), 3 (0.9637 ± 0.0050), and 4 (0.9548 ± 0.0074) than the other models (p < .05). Analysis using layer-integrated gradients showed that the domain-specific BERT model could detect important words, such as disease names in reports.ConclusionsDomain-specific BERT has superiority over the other models in assessing our newly proposed criteria called RIC of head CT radiology reports. The accumulation of similar and further studies of has a potential to contribute to medical safety by preventing missed important findings by clinicians.

Nomogram based on multimodal magnetic resonance combined with B7-H3mRNA for preoperative lymph node prediction in esophagus cancer.

Accurate preoperative prediction of lymph node metastasis (LNM) in esophageal cancer (EC) patients is of crucial clinical significance for treatment planning and prognosis. To develop a clinical radiomics nomogram that can predict the preoperative lymph node (LN) status in EC patients. A total of 32 EC patients confirmed by clinical pathology (who underwent surgical treatment) were included. Real-time fluorescent quantitative reverse transcription-polymerase chain reaction was used to detect the expression of B7-H3 mRNA in EC tissue obtained during preoperative gastroscopy, and its correlation with LNM was analyzed. Radiomics features were extracted from multi-modal magnetic resonance imaging of EC using Pyradiomics in Python. Feature extraction, data dimensionality reduction, and feature selection were performed using XGBoost model and leave-one-out cross-validation. Multivariable logistic regression analysis was used to establish the prediction model, which included radiomics features, LN status from computed tomography (CT) reports, and B7-H3 mRNA expression, represented by a radiomics nomogram. Receiver operating characteristic area under the curve (AUC) and decision curve analysis (DCA) were used to evaluate the predictive performance and clinical application value of the model. The relative expression of B7-H3 mRNA in EC patients with LNM was higher than in those without metastasis, and the difference was statistically significant (P < 0.05). The AUC value in the receiver operating characteristic (ROC) curve was 0.718 (95%CI: 0.528-0.907), with a sensitivity of 0.733 and specificity of 0.706, indicating good diagnostic performance. The individualized clinical prediction nomogram included radiomics features, LN status from CT reports, and B7-H3 mRNA expression. The ROC curve demonstrated good diagnostic value, with an AUC value of 0.765 (95%CI: 0.598-0.931), sensitivity of 0.800, and specificity of 0.706. DCA indicated the practical value of the radiomics nomogram in clinical practice. This study developed a radiomics nomogram that includes radiomics features, LN status from CT reports, and B7-H3 mRNA expression, enabling convenient preoperative individualized prediction of LNM in EC patients.

From technical to understandable: Artificial Intelligence Large Language Models improve the readability of knee radiology reports.

The purpose of this study was to evaluate the effectiveness of an Artificial Intelligence-Large Language Model (AI-LLM) at improving the readability of knee radiology reports. Reports of 100 knee X-rays, 100 knee computed tomography (CT) scans and 100 knee magnetic resonance imaging (MRI) scans were retrieved. The following prompt command was inserted into the AI-LLM: 'Explain this radiology report to a patient in layman's terms in the second person:[Report Text]'. The Flesch-Kincaid reading level (FKRL) score, Flesch reading ease (FRE) score and report length were calculated for the original radiology report and the AI-LLM generated report. Any 'hallucination' or inaccurate text produced by the AI-LLM-generated report was documented. Statistically significant improvements in mean FKRL scores in the AI-LLM generated X-ray report (12.7 ± 1.0-7.2 ± 0.6), CT report (13.4 ± 1.0-7.5 ± 0.5) and MRI report (13.5 ± 0.9-7.5 ± 0.6) were observed. Statistically significant improvements in mean FRE scores in the AI-LLM generated X-ray report (39.5 ± 7.5-76.8 ± 5.1), CT report (27.3 ± 5.9-73.1 ± 5.6) and MRI report (26.8 ± 6.4-73.4 ± 5.0) were observed. Superior FKRL scores and FRE scores were observed in the AI-LLM-generated X-ray report compared to the AI-LLM-generated CT report and MRI report, p < 0.001. The hallucination rates in the AI-LLM generated X-ray report, CT report and MRI report were 2%, 5% and 5%, respectively. This study highlights the promising use of AI-LLMs as an innovative, patient-centred strategy to improve the readability of knee radiology reports. The clinical relevance of this study is that an AI-LLM-generated knee radiology report may enhance patients' understanding of their imaging reports, potentially reducing the responder burden placed on the ordering physicians. However, due to the 'hallucinations' produced by the AI-LLM-generated report, the ordering physician must always engage in a collaborative discussion with the patient regarding both reports and the corresponding images. Level IV.

Abdominal computed tomography scoring systems and experienced radiologists in the radiological diagnosis of small bowel and mesenteric injury

PurposeBlunt bowel and/or mesenteric injury requiring surgery presents a diagnostic challenge. Although computed tomography (CT) imaging is standard following blunt trauma, findings can be nonspecific. Most studies have focused on the diagnostic value of CT findings in identifying significant bowel and/or mesenteric injury (sBMI). Some studies have described scoring systems to assist with diagnosis. Little attention, has been given to radiologist interpretation of CT scans. This study compared the discriminative ability of scoring systems (BIPS and RAPTOR) with radiologist interpretation in identifying sBMI.MethodsWe conducted a retrospective chart review of trauma patients with suspected sBMI. CT images were reviewed in a blinded fashion to calculate BIPS and RAPTOR scores. Sensitivity and specificity were compared between BIPS, RAPTOR, and the admission CT report with respect to identifying sBMI.ResultsOne hundred sixty-two patients were identified, 72 (44%) underwent laparotomy and 43 (26.5%) had sBMI. Sensitivity and specificity were: BIPS 49% and 87%, AUC 0.75 (0.67–0.81), P < 0.001; RAPTOR 46% and 82%, AUC 0.72 (0.64–0.79), P < 0.001; radiologist impression 81% and 71%, AUC 0.82(0.75–0.87), P < 0.001. The discriminative ability of the radiologist impression was higher than RAPTOR (P = 0.04) but not BIPS (P = 0.13). There was not a difference between RAPTOR vs. BIPS (P = 0.55).ConclusionRadiologist interpretation of the admission CT scan was discriminative of sBMI. Although surgical vigilance, including evaluation of the CT images and patient, remains fundamental to early diagnosis, the radiologist’s impression of the CT scan can be used in clinical practice to simplify the approach to patients with abdominal trauma.

Difficulty in the diagnosis of pancreatic cancer based on the initial CT report: A retrospective study.

The role of computed tomography (CT) in the initial diagnosis of pancreatic cancer (PC) is well-known. CT reports made by radiologists are important as not all patients with PC are examined by specialists; however, some cases are not identified based on CT reports. Diagnosis via imaging of PC is sometimes difficult, and the diagnostic rate of PC and other pancreatic diseases can vary across radiologists. This study aimed to examine the diagnostic rate of PC in initial CT reports and the details of cases with diagnostic difficulties. This single-centered, retrospective study collected clinical data of 198 patients with histologically diagnosed PC between January 2018 and April 2022. Out of these contrast-enhanced CT was performed in 192 cases. PC was not reported as the main diagnosis in 18 patients (9.4%; 11 men and 7 women). Among these 18 cases, intrapancreatic mass lesions were detected in 3 (1.6%), indirect findings such as bile/pancreatic duct stenosis or dilation were detected in 5 (2.6%), and no PC-related findings were found in 10 (5.2%). The specialists suspected PC in 15 of these 18 cases based on initial CT reports. 17 cases were confirmed by endoscopic ultrasound-fine needle aspiration and one by biopsy after upper gastrointestinal endoscopy. To improve accuracy of its diagnosis, it is important that specialists provide feedback to diagnostic radiologists regarding the findings they did not report. Endoscopic ultrasound-fine needle aspiration should be performed by specialists when there is clinical information which indicates pancreatic disease of any kind.

Are clinically unimportant findings qualified as benign in lumbar spine imaging reports? A content analysis of plain X-ray, CT and MRI reports.

Lumbar spine diagnostic imaging reports may cause patient and clinician concern when clinically unimportant findings are not explicitly described as benign. Our primary aim was to determine the frequency that common, benign findings are reported in lumbar spine plain X-ray, computed tomography (CT) and magnetic resonance imaging (MRI) reports as either normal for age or likely clinically unimportant. We obtained 600 random de-identified adult lumbar spine imaging reports (200 X-ray, 200 CT and 200 MRI) from a large radiology provider. Only reports requested for low back pain were included. From the report text, one author extracted each finding (e.g., 'broad-based posterior disc bulge') and whether it was present or absent (e.g., no disc bulge) until data saturation was reached, pre-defined as a minimum of 50 reports and no new/similar findings in the last ten reports within each imaging modality. Two authors independently judged whether each finding was likely clinically unimportant or important. For each likely clinicially unimportant finding they also determined if it had been explicitly reported to be benign (expressed as normal, normal for age, benign, clinically unimportant or non-significant). Data saturation was reached after coding 262 reports (80 X-ray, 82 CT, 100 MRI). Across all reports we extracted 3,598 findings. Nearly all reports included at least one clinically unimportant finding (76/80 (95%) X-ray, 80/82 (98%) CT, 99/100 (99%) MRI). Over half of the findings (n = 2,062, 57%; 272 X-Ray, 667 CT, 1123 MRI) were judged likely clinically unimportant. Most likely clinically unimportant findings (90%, n = 1,854) were reported to be present on imaging (rather than absent) and of those only 18% (n = 331) (89 (35%) X-ray, 93 (16%) CT and 149 (15%) MRI) were explicitly reported as benign. Lumbar spine imaging reports frequently include findings unlikely to be clinically important without explicitly qualifying that they are benign.

Infection prevention and control in CT Part 1: An Australian study of roles, responsibilities and practices

IntroductionThe computed tomography (CT) suite presents Infection Prevention and Control (IPC) risks for staff and patients. This is the first of a two-part series on IPC in the CT setting and reports on Australian baseline data related to intravenous contrast media administration and the use of power injectors in the CT suite, including knowledge sources related to the delivery of contrast media. The second part provides insights into CT staff's perceptions of high-risk scenarios for non-adherence to standard or transmission-based precautions. MethodsThe study employed an online survey design, directed at radiographers and radiology nurses working in Australia. The survey included questions relating to CT workplace staffing and equipment, contrast media use and occupational roles, and knowledge sources used for infection control and contrast injectors. ResultsOverall, 160 study participants completed the survey (radiographers: n = 138, 86.3%; nurses: n = 22, 13.7%). Differences were identified between public and private practice. Public hospitals completed more contrast-enhanced scans, and with dual injecting systems, operated and cleaned by radiographers and nurses. Private clinics generally used single-system, power injectors. Radiographers and nurses relied heavily on their colleagues and product guidelines for IPC information. IPC teams were uncommon in private clinics, very common in public departments and 50% of respondents had undertaken IPC training in the last 12 months. ConclusionInsights into use and duties of professionals delivering intravenous contrast media administration using power injectors in the CT suite will help to inform decision-making processes on IPC education strategies and map risk. Implication for practiceFuture research should focus on how radiology workers in CT perceive IPC risk and/or might vary from best practice, which has direct clinical implications for safety if contamination, or incorrect information is routinely applied.

Extracting Clinical Information From Japanese Radiology Reports Using a 2-Stage Deep Learning Approach: Algorithm Development and Validation.

Radiology reports are usually written in a free-text format, which makes it challenging to reuse the reports. For secondary use, we developed a 2-stage deep learning system for extracting clinical information and converting it into a structured format. Our system mainly consists of 2 deep learning modules: entity extraction and relation extraction. For each module, state-of-the-art deep learning models were applied. We trained and evaluated the models using 1040 in-house Japanese computed tomography (CT) reports annotated by medical experts. We also evaluated the performance of the entire pipeline of our system. In addition, the ratio of annotated entities in the reports was measured to validate the coverage of the clinical information with our information model. The microaveraged F1-scores of our best-performing model for entity extraction and relation extraction were 96.1% and 97.4%, respectively. The microaveraged F1-score of the 2-stage system, which is a measure of the performance of the entire pipeline of our system, was 91.9%. Our system showed encouraging results for the conversion of free-text radiology reports into a structured format. The coverage of clinical information in the reports was 96.2% (6595/6853). Our 2-stage deep system can extract clinical information from chest and abdomen CT reports accurately and comprehensively.

Weakly guided attention model with hierarchical interaction for brain CT report generation

Brain Computed Tomography (CT) report generation, which aims to assist radiologists in diagnosing cerebrovascular diseases efficiently, is challenging in feature representation for dozens of images and language descriptions with several sentences. Existing report generation methods have achieved significant achievement based on the encoder–decoder framework and attention mechanism. However, current research has limitations in solving the many-to-many alignment between the multi-images of Brain CT imaging and the multi-sentences of Brain CT report, and fails to attend to critical images and lesion areas, resulting in inaccurate descriptions. In this paper, we propose a novel Weakly Guided Attention Model with Hierarchical Interaction, named WGAM-HI, to improve Brain CT report generation. Specifically, WGAM-HI conducts many-to-many matching for multiple visual images and semantic sentences via a hierarchical interaction framework with a two-layer attention model and a two-layer report generator. In addition, two weakly guided mechanisms are proposed to facilitate the attention model to focus more on important images and lesion areas under the guidance of pathological events and Gradient-weighted Class Activation Mapping (Grad-CAM) respectively. The pathological event acts as a bridge between the essential serial images and the corresponding sentence, and the Grad-CAM bridges the lesion areas and pathology words. Therefore, under the hierarchical interaction with the weakly guided attention model, the report generator generates more accurate words and sentences. Experiments on the Brain CT dataset demonstrate the effectiveness of WGAM-HI in attending to important images and lesion areas gradually, and generating more accurate reports.

Index CT-Based Scoring Systems in Operative Blunt Bowel and Mesenteric Injury Identification

IntroductionDetermining the need for surgical management of blunt bowel and mesenteric injury (BBMI) remains a clinical challenge. The Faget score and Bowel Injury Prediction Score (BIPS) have been suggested to address this issue. Their efficacy in determining the need for surgery was examined. MethodsA retrospective review of all adult blunt trauma patients hospitalized at a level 1 trauma center between January 2009 and August 2019 who had small bowel, colon, and/or mesenteric injury was conducted. We further analyzed those who underwent preoperative computed tomography (CT) scanning at our institution. Final index CT reports were retrospectively reviewed to calculate the Faget and BIPS CT scores. All images were also independently reviewed by an attending radiologist to determine the BIPS CT score. ResultsDuring the study period, 14,897 blunt trauma patients were hospitalized, of which 91 had BBMI. Of these, 62 met inclusion criteria. Among patients previously identified as having BBMI in the registry, the retrospectively applied Faget score had a sensitivity of 39.1%, specificity of 81.2%, positive predictive value (PPV) of 85.7%, and negative predictive value (NPV) of 31.7% in identifying patients with operative BBMI. The retrospectively applied BIPS score had a sensitivity of 47.8%, specificity of 87.5%, PPV of 91.7%, and NPV of 36.8% in this cohort. When CT images were reviewed by an attending radiologist using the BIPS criteria, sensitivity was 56.5%, specificity 93.7%, PPV 96.3%, and NPV 42.8%. ConclusionsExisting BBMI scoring systems had limited sensitivity but excellent PPV in predicting the need for operative intervention for BBMI. Attending radiologist review of CT images using the BIPS scoring system demonstrated improved accuracy as opposed to retrospective application of the BIPS score to radiology reports.