Automating radiology report generation can significantly alleviate radiologists’ workloads. Previous research has primarily focused on realizing highly concise observations while neglecting the precise attributes that determine the severity of diseases (e.g., small pleural effusion). Since incorrect attributes will lead to imprecise radiology reports, strengthening the generation process with precise attribute modeling becomes necessary. Additionally, the temporal information contained in the historical records, which is crucial in evaluating a patient’s current condition (e.g., heart size is unchanged), has also been largely disregarded. To address these issues, we propose RECAP, which generates precise and accurate radiology reports via dynamic disease progression reasoning. Specifically, RECAP first predicts the observations and progressions (i.e., spatiotemporal information) given two consecutive radiographs. It then combines the historical records, spatiotemporal information, and radiographs for report generation, where a disease progression graph and dynamic progression reasoning mechanism are devised to accurately select the attributes of each observation and progression. Extensive experiments on two publicly available datasets demonstrate the effectiveness of our model.

RECAP: Towards Precise Radiology Report Generation via Dynamic Disease Progression Reasoning

The recent success of general-domain large language models (LLMs) has significantly changed the natural language processing paradigm towards a unified foundation model across domains and applications. In this paper, we focus on assessing the performance of GPT-4, the most capable LLM so far, on the text-based applications for radiology reports, comparing against state-of-the-art (SOTA) radiology-specific models. Exploring various prompting strategies, we evaluated GPT-4 on a diverse range of common radiology tasks and we found GPT-4 either outperforms or is on par with current SOTA radiology models. With zero-shot prompting, GPT-4 already obtains substantial gains 
($\approx$ 10\% absolute improvement) over radiology models in temporal sentence similarity classification (accuracy) and natural language inference ($F_1$). For tasks that require learning dataset-specific style or schema (e.g. findings summarisation), GPT-4 improves with example-based prompting and matches supervised SOTA. Our extensive error analysis with a board-certified radiologist shows GPT-4 has a sufficient level of radiology knowledge with only occasional errors in complex context that require nuanced domain knowledge. For findings summarisation, GPT-4 outputs are found to be overall comparable with existing manually-written impressions.

Exploring the Boundaries of GPT-4 in Radiology | VIDEO

Automatically generated reports from medical images promise to improve the workflow of radiologists. Existing methods consider an image-to-report modeling task by directly generating a fully-fledged report from an image. However, this conflates the content of the report (e.g., findings and their attributes) with its style (e.g., format and choice of words), which can lead to clinically inaccurate reports. To address this, we propose a two-step approach for radiology report generation. First, we extract the content from an image; then, we verbalize the extracted content into a report that matches the style of a specific radiologist. For this, we leverage RadGraph---a graph representation of reports---together with large language models (LLMs). In our quantitative evaluations, we find that our approach leads to beneficial performance. Our human evaluation with clinical raters highlights that the AI-generated reports are indistinguishably tailored to the style of individual radiologist despite leveraging only a few examples as context.

Style-Aware Radiology Report Generation with RadGraph and Few-Shot Prompting

**Scoring is only possible in the semifinals hall**

**Abstract:**



Analyzing the Quality of Reporting of AI Studies for Hip Fracture Detection

Background:
Chordae tendineae rupture (CTR) has been linked to several factors as hypertension, rheumatic fever, and ischemic heart disease. Here we report a patient with CTR during COVID-19 infection.

Case
A 79-year-old man with non-stenotic coronary artery disease, ischemic cardiomyopathy with diastolic dysfunction, rheumatic heart disease with chronic moderate mitral regurgitation, pulmonary hypertension, paroxysmal atrial fibrillation on apixaban, hypertension, familial hypercholesterolemia, and obstructive sleep apnea presented with progressive worsening of dyspnea, palpitations, orthopnea, and lightheadedness for 6 months duration with acute deterioration 2 hours before admission.

He was hemodynamically stable but was hypoxic with saturation at 89%, requiring 2L O2 by nasal cannula. On examination, a holosystolic murmur was auscultated at the cardiac apex. The rest of examination was normal.

EKG showed sinus rhythm at 73 BPM without ST changes. Laboratory investigations were within normal limits except for Brain natriuretic peptide, which was 3160 pg/mL. He tested positive for COVID-19. Chest x-ray revealed patchy perihilar, right lower lung zone opacities, and a minor right-sided pleural effusion. Transesophageal echocardiography showed moderate to severe mitral valve regurgitation, left ventricular hypertrophy with normal left ventricular ejection fraction at 60%. Records of previous imaging were not available.

Upon cardiac catheterization, high-grade proximal stenosis was detected in the left anterior descending artery and ostial small circumflex; flail posterior mitral valve leaflet with ruptured chordae tendineae, redundant and hypermobile anterior mitral leaflet and myxomatous degeneration of the mitral valve was noted. He underwent mitral valve replacement with mosaic porcine bioprosthesis, single-vessel bypass grafting in the left anterior descending artery with left internal mammary graft, left-sided maze atrial ablation atricure cryoprobe, and Mitra Clip placement at the base of left atrial appendage after a week of clinical improvement with diuresis, nirmatrelvir/ritonavir, and holding apixaban. Apixaban was not restarted postoperatively due to poor general health and high fall risk, and he was discharged to rehabilitation.

Discussion
Mitral regurgitation (MR) affects 9.3% of Americans 75 and older. Primary MR is caused by mitral valve prolapse from CTR. CTR can be of primary or secondary origin. Hypertension is determined to be an independent predictor of primary CTR. Acute rheumatic fever, osteogenesis imperfecta, ischemic heart disease, chronic rheumatic valvular disease, and bacterial endocarditis are causes of secondary CTR. Myxomatous degeneration is underreported since the microscopic investigation is not routinely performed. The posterior chorda is more commonly affected than the anterior chorda. Mild mitral CTR often entails the separation of a single chorda, resulting in limited hemodynamic sequelae requiring no treatment. Multiple chordae ruptures may cause moderate-to-severe MR requiring surgical intervention. The course of the presentation can be acute or progressive. MR severity grading influences clinical decisions, prognosis, and treatment plans. Echocardiography is the most routinely conducted noninvasive MR test.
In literature, COVID-19 infection is linked to cardiovascular complications as myocarditis, myocardial infarction, and cardiogenic shock. At this juncture, there is not enough evidence to attribute the development of CTR to COVID-19 infection, although a causal association is possible.

Conclusion
This case highlights a complex presentation of Chordae tendineae rupture with concurrent Covid-19 infection.

Chordae Tendineae Rupture in Chronic Mitral Regurgitation and COVID-19 Infection


Tumor lysis syndrome (TLS) occurs when tumor cells lyse and empty their contents into the bloodstream. TLS can occur in response to therapy or spontaneously, and it is most commonly seen in high-grade lymphomas, acute leukemias, and other rapidly proliferating tumors. For multiple myeloma (MM), the risk of RLS is less than one percent, and the risk of spontaneous TLS is even rarer. Therefore, we present a rare case of spontaneous TLS in MM. This case met the 2004 Cairo-Bishop laboratory criterion as well as the clinical TLS criterion for diagnosis and was treated with rasburicase as per guideline recommendations. We hope to educate others on such a presentation especially given the emergent nature of treatment and the increasing incidence of this disease.

A Rare Case of Spontaneous Tumor Lysis Syndrome in Multiple Myeloma

Background
Renal cell carcinoma (RCC) is a highly heterogenous cancer that typically relies on biopsy to identify the specific tumor subtype and identify high-grade histology. However, accuracy and reliability of biopsy is sometimes limited by tumor heterogeneity. The use of 3D printing to improve the accuracy of this process has been previously described for prostate and renal cancers. Preoperative imaging is used to 3D print “negative” molds matching the contour of the surgically removed tumor and provide guiding lines along which to cut precise slices that correspond with imaging slices. Our objective was to design a novel technique combining negative molds with additional 3D-printed overlays, or “positive” molds that would fit onto the pathology slices and have cutouts to denote regions of interest previously identified on imaging, further facilitating the sectioning process and correlating pathology findings to imaging in RCC.

Methods
This was a single-center, IRB-approved, prospective study in patients with RCC who planned to undergo partial or radical nephrectomy. Preoperative images were transferred to a viewing software (3D Slicer), and their data was used to generate “positive” and “negative” 3D -printed molds. Following nephrectomy, kidneys were sectioned by a study pathologist and study radiologist using the 3D-printed molds. Pathology slides were then directly compared to CT by a radiologist using 3D Slicer and Docker to determine degree of pathology-to-radiology correlation. 

Results
11 total 3D-printed molds were constructed. Mold failure was defined by cases where 3D-printed molds could not be used throughout the entire sectioning process, or where lesions on pathology did not align with preoperative imaging. Mold success was defined by cases with complete mold-assisted sectioning and correlation of pathology to radiology images. The first 5 molds produced failed, either due to tumor growth between imaging and sectioning (n=3) or perirenal fat (n=2). Improvements in mold design using a bivalving technique were achieved following discussion between the study urologist, pathologist, and radiologist, and the subsequent 6 molds produced were successful.

Conclusion
In this study, we aimed to develop a novel technique for designing patient-specific renal molds to improve the diagnostic accuracy of tumor grading in renal cell carcinoma (RCC), which we achieved following multidisciplinary discussion and process revision. This is the first study to describe such an approach using negative and positive 3D-printed molds, and further research is warranted to validate the efficacy and efficiency of our proposed methodology in order to better standardize tumor sectioning and advance the diagnostic process for RCC.


3D Printed Molds for Pathology Sectioning of Renal Masses: Enabling Radiology-Pathology Correlation

Background
Myasthenia gravis is classically described in the literature as an autoimmune disorder with antibodies directed at the acetylcholine receptor on the motor end plate. This autoimmune destruction of these key areas of the interface between the nervous system and the musculoskeletal system leads to weakness and fatigue in patients. Rarely in the literature has this been described as a molecular mimicry event secondary to viral illnesses like that of COVID-19. 

Case Report
Herein the authors describe the case of a 71-year-old right-handed female who presented with key symptoms of dyspnea, drooping eye lids, difficulty lifting her neck, dysphagia, slurring of speech, bilateral shoulder weakness, and needing to change eating positions due to discomfort of her shoulders and arms. The patient endorsed her eyelid drooping worsened in the later part of the day such that she had to tape her eyelids open in order to see. The patient underwent ice testing, trial of pyridostigmine, was discharged with a prescription for prednisone and pyridostigmine, set up with outpatient follow up, and was given a working diagnosis of COVID-19 induced myasthenia gravis.

Discussion
This case is important as it highlights three key lessons of importance within the literature on myasthenia gravis. First is the transient worsening of a patient when given corticosteroids. In high doses, steroids can worsen or send a person into myasthenic crisis which is the reason that they should not be utilized as primary treatment in crisis. Second is the etiology of myasthenia gravis, often there is not found an antecedent event in patients, however, the history of this case made it clear that this was likely a molecular mimicry event secondary to a COVID-19 infection. Thus, the authors desire to remind the readership of the importance of a thorough history and physical examination. Lastly, as a part of the diagnostic work up, the clinicians utilized an older physical examination maneuver known as the ice test. When utilized, this test has a high negative predictive value and therefore still has a place in bedside diagnostics when ruling out myasthenia gravis.

"It's been downhill since COVID": Myasthenia Gravis secondary to COVID Infection

Background
Rural EMS agencies have long struggled to financially maintain operations because of low call volumes and higher operating costs, and closures have only increased as higher portions of the rural population are uninsured or covered under Medicare and Medicaid. Similarly, community health workers (CHWs) do not have enough volume in rural areas to be financially sustainable. Community paramedicine (CP) models, which combine elements of EMS and CHWs, have demonstrated remarkable success in improving care and reducing costs. However, these models have had limited uptake in rural areas due to reimbursement and regulatory challenges and the models’ emphasis on paramedics rather than inexpensive emergency medical technicians-basic (EMT-Bs) or advanced emergency medical technicians (AEMTs).

Methods
We conducted a narrative literature review to assess the feasibility of dual-certifying EMT-Bs and AEMTs as CHWs to improve the financial sustainability of rural EMS agencies and improve healthcare access. We used Pubmed to search for both original articles and reviews on rural CP, and Google was used to identify gray literature relating to the regulatory, financial, and governance structures of CP and CHWs. Both scholarly and gray literature were only included if they included studies in the USA and were within the last 10 years. 

Results
The scope of practice and training of CP programs varied widely, but most programs involved care coordination, health assessments, health education and promotion, and prevention of ED visits and rehospitalizations. EMT-Bs and AEMTs are in 29% and 10% of CP programs, respectively, even though the majority of rural EMS personnel are non-paramedics. Five states provide Medicaid reimbursement specifically for CP services, and fourteen states provide EMS reimbursement for treatment without transport, but not specifically CP services. Medicare only pays for some CP services under the experimental ET3 model. Conversely, at least 29 state Medicaid agencies reimburse for CHW services, and 21 states have CHW certification programs with another 14 states currently developing them. Medicare also pays for CHW services as of 2021. 

Conclusion
Training rural EMT-Bs/AEMTs to become certified CHWs may be a viable method of increasing EMS agencies’ income by increasing their reimbursable patient encounters in states that reimburse CHWs. This would enable rural EMS providers to avoid the regulatory and reimbursement challenges of other CP programs while increasing the financial sustainability of their agency. While the CHW and EMT-B scopes of practice will limit the extent of care they can provide more than paramedic-based programs, they could significantly expand rural access to both EMS and CHW services with appropriate physician oversight.

Rethinking EMS in the Fight for Rural Health Equity

BACKGROUND:
Rheumatoid arthritis (RA) is a chronic autoimmune disease impacting about 1.3 million Americans, with women much more likely to be affected than men. The development of this systemic inflammatory disease is influenced by both genetic and environmental factors. Certain environmental factors, such as diet, can be more readily controlled by individuals, and therefore may serve as more attainable therapeutic targets for preventing and mitigating RA and its symptoms. 

Various foods and nutrients are associated with inflammatory biomarkers1,2,3,17, and thus have been linked with increased or decreased risk of RA. However, these associations have been inconsistent across studies likely due to overall systematic bias, measurement error associated with individual food groups and nutrients, and limited investigations evaluating the relationship between overall dietary patterns and RA in larger sample sizes4,5,6,12. Furthermore, few studies that have examined the impact of overall dietary patterns on RA but have had conflicting results and/or generally involved smaller sample sizes. 

OBJECTIVE:
Our objective is to evaluate the association between dietary quality, as measured by the 2015 Healthy Eating Index (HEI-2015), and risk of incident rheumatoid arthritis (RA) in postmenopausal women.

METHODS: 
109,591 postmenopausal women without baseline RA were prospectively followed as part of the Women’s Health Initiative (WHI) for an average of 8.1 years. Baseline diet was measured using a food frequency questionnaire (FFQ) and diet quality was evaluated using the HEI-2015. Multivariable Cox proportional regression analysis was performed evaluating the association of diet quality with self-reported physician diagnosed RA after adjusting for age, race, education status, income, and body mass index (BMI). 

RESULTS:
During 1,241,532 person-years of follow-up, 5,823 incident RA cases were identified. Compared to quartile 1, quartiles 2, 3, and 4 of the HEI-2015 scores were associated with a reduced RA risk of 3%, 13%, and 18%, respectively (p-trend <0.001). A higher score for total fruits (p-trend=.006), whole fruits (p-trend<.001), total vegetables (p-trend=.004), dark green vegetables (p-trend<.001), whole grain (p-trend=.004), total dairy (p-trend=.004), and seafood/plant protein (p-trend=.015) consumption was significantly associated with lower rates of incident RA independent of age and confounders. Conversely, a higher score for consumption of percent saturated fat (p-trend=.001) was significantly associated with higher rates of incident RA (Table 3).

CONCLUSIONS:
A higher quality diet rich in fruits, vegetables, whole grain, dairy, and seafood/plant protein, is associated with a lower risk of developing RA in postmenopausal women. Meanwhile, a less healthy diet with higher intake of percent saturated fat is associated with higher risk of RA.

Association of Diet Quality with Risk of Incident Rheumatoid Arthritis in the Women’s Health Initiative

Background
Geminus Head Start is a federal program that works with children under five years old from low-income communities to promote school readiness. The organization in Northwest Indiana provided anecdotal data on delayed social and emotional development of children during the pandemic. Studies have shown that adverse childhood development can have a severe psychological and physiological impact on health outcomes into adulthood. However, there has been a lack of research on quantifying the extent of damage from the pandemic on our Head Start children, who mainly identify as Black/African American. 

Methods
Geminus Head Start provided demographic data including race, postal codes and DECA for over 4650 students from 2019-2023. DECA, Devereux Early Childhood Assessment, is used by educators to evaluate children’s social emotional competence for early intervention. In this study, we are investigating the DECA categories, including initiative, self control, and attachment/relationships. We analyzed DECA scores via ANOVA along with listening into the teacher focus groups. 

Results
Students had increased attachment/relationship scores from the 2019-2020 school year to the 2020-2021 school year (36.69 to 44.59). However, following COVID-19 to the 2021-2022 and 2022-2023 school years, these scores dropped to 23.51. When looking at the second protective factor, self regulation, the opposite trend was found. The scores went from 18.79 in 2019-2020 down to 14.68 in 2020-2021 and eventually up to 23.51 in 2021-2022.

Conclusion

Like most institutions, the pandemic has created a time of isolation and instability for our children and their families. In particular, these children have not been able to access adequate conditions to develop emotional and social maturity. Lacking this development can lead to several negative health outcomes as it may impact mental health and the ability to create healthy supportive networks.


Retrospective Analysis of COVID-19 Impact on Social & Emotional Development in Children of Low-Income Communities from Head Start 


Numerous studies have established a strong connection between the heart and brain, especially in response to stressful stimuli. One condition, Takotsubo cardiomyopathy (TC), is characterized by a temporarily weakened left ventricle apex resulting from excess catecholamines and microvascular dysfunction. Another different condition, posterior reversible encephalopathy syndrome (PRES), is associated with acute neurological changes and changes in the occipital lobe. These two conditions share similarities despite presenting different clinical features and requiring distinct diagnostic workups and treatments.

Case Description
A 49-year-old woman with a past medical history of hypertension and migraines presented after a witnessed seizure following a few hours of worsening headaches. She reported high stress, with her husband diagnosed with blood cancer and her dog recently passing away. Her blood pressure was mildly elevated at 156/92. Initially, she appeared to have right-sided muscle weakness, which resolved. A CT brain scan demonstrated occipital lobe changes, suggesting PRES. A CTA brain revealed no major vascular blockages, and an EEG revealed no epileptiform activity. The next day, she reported substernal chest pain and an ECG revealed new dynamic T-wave inversions in the anterior leads. Her troponins were elevated (peaking at 2222 ng/ml), and a TTE showed apical wall abnormalities. A subsequent cardiac catheterization revealed no coronary occlusion, suggesting TC. The patient improved clinically with conservative therapy during the hospitalization and was eventually discharged home.

Discussion
The "heart and brain connection" was previously identified in patients with acute neurologic insults. PRES is a rare neurologic entity; manifestations include headaches, focal neurologic deficits, seizures, vision loss, or encephalopathy. PRES is believed to involve cerebral blood flow alterations and brain autoregulation dysfunction. Imaging findings include vasogenic edema involving the posterior regions of the brain. It can be seen in hypertensive emergencies, renal failure, autoimmune diseases, and sometimes medication-induced flares. PRES is usually reversible. However, a recent small study showed a mortality rate of 4% and residual damage of 5.5%. TC, aka apical ballooning syndrome, previously known as stress-induced cardiomyopathy, has a 1–3% prevalence in postmenopausal females and 5–6% in those with ACS. It is usually triggered by underlying physical or emotional stress, as reported in 60–70% of cases. Studies have shown that 2% of ACS is due to TC. Hence, it is considered one of the ACS-mimicking diseases. Usually present with chest pain or new onset heart failure with characteristic echo findings of apical balloon shaping. The cardiac catheter is usually negative for any obstruction. 
Based on the available data, there is a possible association between PRES and TC. One study reported a prevalence of 2.7%, and around 14 cases were found in the literature, eight of which are in PubMed journals. It is noteworthy that 13 of the 14 cases involved postmenopausal women over the age of 45 years old. While these conditions can be serious, most patients have a good prognosis with appropriate treatment and support. Further research is needed to understand better the connection between these two conditions and their long-term implications. Treatment approaches may vary depending on the patient's circumstances, so seeking a multidisciplinary approach to care is important.

Conclusion
Healthcare providers should keep a high clinical index of suspicion for potential heart and neurological issues like PRES and TC in postmenopausal women with acute coronary syndrome. Neglecting or ignoring TC can lead to a worse prognosis. It is worth noting that neurological deficits associated with TC may indicate PRES. Further research must be conducted to understand and treat these serious illnesses.



Underline digital video library

RECAP: Towards Precise Radiology Report Generation via Dynamic Disease Progression Reasoning

Exploring the Boundaries of GPT-4 in Radiology | VIDEO

Style-Aware Radiology Report Generation with RadGraph and Few-Shot Prompting

Analyzing the Quality of Reporting of AI Studies for Hip Fracture Detection

Chordae Tendineae Rupture in Chronic Mitral Regurgitation and COVID-19 Infection

A Rare Case of Spontaneous Tumor Lysis Syndrome in Multiple Myeloma

3D Printed Molds for Pathology Sectioning of Renal Masses: Enabling Radiology-Pathology Correlation

"It's been downhill since COVID": Myasthenia Gravis secondary to COVID Infection

Rethinking EMS in the Fight for Rural Health Equity

Association of Diet Quality with Risk of Incident Rheumatoid Arthritis in the Women’s Health Initiative

Retrospective Analysis of COVID-19 Impact on Social & Emotional Development in Children of Low-Income Communities from Head Start

A Rare Combination: A Case of a Patient with Takotsubo Cardiomyopathy and Posterior Reversible Encephalopathy Syndrome (PRES)

Stay up to date with the latest Underline news!

PRESENTATIONS

CONFERENCES

COMPANY

RESOURCES