Graphical abstract

Introduction

Background/rationale

Chest radiographs are among the most frequently requested and interpreted imaging studies in clinical practice, yet interpretation errors remain common among junior physicians and medical students [1,2]. Despite the critical role of chest radiographs in diagnosing respiratory and cardiac conditions, formal training in chest radiology is often insufficient, and medical students report low competence and confidence [3]. Online learning platforms offer flexibility, repetition, and accessibility, which are important attributes in clinical education [4]. Within these, formative self-assessment tools allow learners to practice, reflect, and receive feedback without academic pressure [5,6]. However, few resources focus specifically on practical radiographic interpretation in a realistic clinical context [7,8].

Objectives

The aim of this study was to describe the development, implementation, and educational use of Radiotorax.es, a freely accessible web platform that enables formative self-assessment in interpreting chest X-rays, and to report on user activity over 11 years.

Ethics statement

Radiographic cases were anonymized before being included in the system. All data were processed anonymously, in compliance with data protection regulations. This study was approved by the Ethics Committee for Experimentation at the University of Málaga under reference 140-2022-H, on 18 January 2023.

Content and implementation of the web-based resource

Purpose

Radiotorax.es addresses an educational need by offering a realistic, low-pressure environment to practice and assess chest X-ray interpretation. Designed to complement formal training, it promotes reflective learning through repetition, feedback, and progress tracking.

Cases

Four hundred cases were selected from the picture archiving and communication system of the Radiology Department at Virgen de la Victoria University Hospital (Málaga, Spain): 80 normal (20%) and 320 pathological (80%), covering a wide range of findings—bone, pulmonary, pleural, cardiac, mediastinal, and abdominal. These were further classified into 215 frequent (54%), 45 subtle (11%), and 60 severe cases (15%) requiring urgent intervention.

All cases were anonymized before inclusion. Images were stored in low-compression, full-resolution JPEG format (2992×2997 px, 96 dpi, 24-bit). Each included one posteroanterior (PA) view or both PA and lateral projections. Original reports were extracted from the hospital’s Radiology Information System.

Website architecture

The platform was developed using PHP ver. 5.5 (PHP Group), JavaScript ver. 5.1 (ECMA International), HTML (W3C), and CSS ver. 2.0 (W3C), with MySQL ver. 5.0 (https://www.mysql.com/) managing the database. The application is modular (Fig. 1) and hosted at www.radiotorax.es.

User registration

Users register with their name, email, gender, professional category (student, resident, specialist), training year, specialty (if applicable), and prior radiology training (yes/no). All data are confidential and accessible only to the user or system administrator.

Self-assessment

Registered users may complete 2 training rounds (3 cases each) to familiarize themselves with the platform. Full assessments simulate clinical reporting sessions: 20 chest X-rays interpreted in 60 minutes. Each round includes 7 normal X-rays, 7 X-rays showing frequent pathologies, 3 severe cases, and 3 subtle cases, presented with patient age, sex, and clinical data when available. Some cases lack clinical context, reflecting real-life scenarios.

Fig. 2 outlines the session flow. Each case presents clinical and imaging data, followed by a report entry (Supplement 1). After all 20 cases, users compare their reports with expert versions in a structured self-assessment, answering: (1) was your report correct? (yes, no); (2) did you describe all findings? (yes, no); (3) did you detect the main finding? (yes, no); (4) perceived difficulty (very low, low, medium, high, very high); and (5) report quality (1=poor, 10=excellent). A short demonstration video clip showing the user workflow during a self-assessment session is available in Supplement 2.

Record of self-assessments

All reports and results are stored in a personal database. After each session, a summary displays the date, time spent, number of correct reports, completeness of descriptions, identification of main findings, average difficulty, and report score. A PDF is generated for each session, containing all data and links to full-resolution images (Supplement 3). Users can access their session history and download reports for review or feedback.

Intended audience

The tool targets medical students, residents, and physicians seeking to improve or maintain chest X-ray interpretation skills.

Access and educational integration

Radiotorax.es is a fully online, asynchronous tool accessible via any modern browser. As a nonprofit project, it is freely available to the academic and medical community, removing financial and institutional barriers.

It supports self-directed learning, adapts to all training levels, and has been integrated into the undergraduate curriculum at the University of Málaga (UMA), as well as postgraduate and continuing medical education.

Evaluation and impact

Assessment methods

The tool includes a structured self-evaluation after each case that assesses descriptive accuracy, diagnostic success, case difficulty, and report quality. Each session generates a downloadable PDF for instructor feedback or monitoring progress. The assessment combines yes/no items and Likert-scales, supporting structured reflection.

User activity

User registration data were extracted from the Radiotorax.es database (Dataset 1). From May 30, 2011, to June 1, 2022, Radiotorax.es registered 14,389 users, 7,726 of whom completed at least one self-assessment: 4,268 women (55.2%) and 3,458 men (44.8%). Most users were from Spain (5,890; 76.2%), followed by Argentina (487), Mexico (410), and Colombia (222), among 49 other countries (Fig. 3).

In Spain, 2,312 users (39.2%) were medical students from UMA, where the platform is part of formal radiology training. The remaining 3,578 users were distributed across Spanish provinces, notably Las Palmas (601), Madrid (247), and Barcelona (214) (Fig. 4).

By profession, 5,583 users (75.8%) were medical students, 1,174 (15.2%) residents, and 699 (9.0%) specialists. Most (57.1%) had no prior radiology training. Users completed 18,667 self-assessments, averaging 2.4 per person (Table 1). Among students, 39.5% were from UMA and 60.5% from other institutions, completing 11,491 and 4,296 sessions (average: 5.0 and 1.4 sessions), respectively.

Supplement 4 summarizes users’ distribution by training year—highlighting, for example, 3,562 third-year students (60.9%) and 627 first-year residents (53.4%)—and their medical specialties. It also reports average self-assessment scores by user type and question.

Potential impact

Radiotorax.es has proven highly versatile in education. Since 2011, it has been regularly used in the third and sixth years of the medical program at the University of Málaga, with over 2,300 students enrolled. As a free-access resource, it supports diverse uses by learners and instructors.

Personal self-assessment

Radiotorax.es allows users to independently assess their chest X-ray interpretation skills at any time through 20-case rounds. Generated PDFs help track progress—for example, before and after a course or clinical rotation.

Instructor-guided evaluation

Educators can assign self-assessments at key points in training. PDFs serve as documentation and support performance review, enabling instructors and learners to monitor progress.

Tutorial support

In tutoring sessions, PDFs can be used to review cases, resolve doubts, and discuss interpretive errors. High-resolution image links support detailed feedback based on the learner’s prior work.

Peer review

Students may review a peer’s self-assessment, anonymously or not. This fosters diagnostic reasoning for the reviewer and provides constructive feedback for the author [8].

Expert feedback

Faculty can use PDFs for secondary correction and benchmarking, helping learners evaluate their self-assessment accuracy against expert standards.

Multi-center or longitudinal studies

Radiotorax.es facilitates large-scale educational research. Its flexible design allows integration across educational levels and institutions, supporting studies on learning patterns and diagnostic development.

Discussion

This article describes the development and implementation of Radiotorax.es, a web-based tool for formative self-assessment in chest X-ray interpretation. Over 11 years, more than 14,000 users registered, with over half completing assessment sessions. Its sustained use among students, residents, and clinicians highlights demand for accessible, practical radiology training.

Chest X-ray interpretation remains a core yet challenging skill [9]. Studies have revealed persistent gaps in competence and confidence among students and junior doctors, even when identifying normal or subtle findings [1-3]. Radiotorax.es addresses this through structured cases that mirror interpretive practices advocated since the mid-20th century [10].

Formative self-assessment fosters self-awareness, motivation, and deeper learning by reducing the pressure of summative evaluation [5,6]. Radiotorax.es supports this through expert feedback, session summaries, and multiple applications, including personal tracking, peer review, and instructor-led use.

As a non-profit, open-access tool based on real cases, Radiotorax.es enables both curriculum integration and independent learning, and facilitates collaborative research across diverse educational settings. With non-radiologist clinicians increasingly interpreting imaging in urgent care, such tools are especially relevant today. Although commercial platforms for radiology training exist, Radiotorax.es offers a freely accessible, formative self-assessment tool focused specifically on chest X-ray interpretation.

Limitations

This presentation of educational material is descriptive. Direct measurements of learning outcomes are a matter of ongoing research. Future studies will assess the educational effectiveness of Radiotorax.es through pre/post testing and correlation with clinical performance. Given its reliance on self-assessment, there is potential for bias, including overestimation and underestimation. Future work may address these issues using objective performance metrics, external expert evaluations, or generative artificial intelligence (AI) tools to analyze learners’ written reports. Finally, the current availability of the platform only in Spanish may limit generalizability to non-Spanish-speaking settings.

Conclusion

Radiotorax.es is a well-established, widely used web-based platform that enhances radiology education through formative self-assessment. It promotes reflective practice, longitudinal learning, faculty-led activities, and adapts to various educational levels and contexts. Future developments may include expanding the case library, adding English-language support, improving analytic tools, updating the technological platform to implement responsive design for better mobile accessibility, and exploring the relationship between self-assessment outcomes, self-perception, and clinical performance. The integration of AI-based tools may also improve feedback and training, with generative AI serving as an external evaluator of written reports to identify and quantify overestimation and underestimation biases.

Supplementary materials

Supplement 2. Radiotorax_Self_Assessment_Video_Clip.

Fig. 1.

Flowchart of the Radiotorax.es application. First, users must register by completing the requested data. They can complete up to 2 training rounds of 3 cases and as many self-assessment rounds of 20 cases as they wish. Each round generates a PDF file with the detailed results. Users can access their self-assessment history and download the corresponding PDF files.

Fig. 2.

Flowchart of the interpretive self-assessment round of 20 cases. Once started, the images and clinical data of each case are displayed sequentially, with a text box where the user must enter the corresponding report. Subsequently, the 20 cases are presented again with the user report together with the expert report and the options to complete the self-assessment data. When finished, the screen displays a summary and a link to the created PDF file.

Fig. 3.

Distribution by country of Radiotorax.es users who completed at least one self-assessment. The orange circles are in proportion to the number of users. The light blue circles correspond to those countries with less than 10 users. The 2,312 users corresponding to medical students at the University of Malaga are excluded. NA, not answered.

Fig. 4.

Distribution by province of users in Spain who completed at least one self-assessment with Radiotorax.es. The orange circles are proportional to the number of users. Provinces with more than 100 users are specifically identified. Light blue circles correspond to those provinces with less than 20 users. The 2,312 users corresponding to medical students at the University of Malaga are excluded. NA, not answered.

References

• 1. Eisen LA, Berger JS, Hegde A, Schneider RF. Competency in chest radiography: a comparison of medical students, residents, and fellows. J Gen Intern Med 2006;21:460-465. https://doi.org/10.1111/j.1525-1497.2006.00427.x ArticlePubMedPMC
• 2. Potchen EJ, Cooper TG, Sierra AE, Aben GR, Potchen MJ, Potter MG, Siebert JE. Measuring performance in chest radiography. Radiology 2000;217:456-459. https://doi.org/10.1148/radiology.217.2.r00nv14456 ArticlePubMed
• 3. Jeffrey DR, Goddard PR, Callaway MP, Greenwood R. Chest radiograph interpretation by medical students. Clin Radiol 2003;58:478-481. https://doi.org/10.1016/s0009-9260(03)00113-2 ArticlePubMed
• 4. Sendra Portero F, Munoz Nunez CF. Online training tools in radiology. Radiologia 2011;53:498-505. https://doi.org/10.1016/j.rx.2011.02.011 ArticlePubMed
• 5. Khan KS, Davies DA, Gupta JK. Formative self-assessment using multiple true-false questions on the Internet: feedback according to confidence about correct knowledge. Med Teach 2001;23:158-163. https://doi.org/10.1080/01421590031075 ArticlePubMed
• 6. Kaminski JJ, Franz A, Holzhausen Y, Wedenig HH, Sporn F, Peters H. Establishing TELLme: an online platform for self-directed, formative assessment of knowledge acquisition across an undergraduate medical curriculum. Med Teach 2025;47:696-704. https://doi.org/10.1080/0142159X.2024.2362896 ArticlePubMed
• 7. Chew FS. Radiology of the hands: review and self-assessment module. AJR Am J Roentgenol 2005;1846 Suppl):S157-S168. https://doi.org/10.2214/ajr.184.6_supplement.0184s157 ArticlePubMed
• 8. McEvoy FJ, McEvoy PM, Svalastoga EL. Web-based teaching tool incorporating peer assessment and self-assessment: example of aligned teaching. AJR Am J Roentgenol 2010;194:W56-W59. https://doi.org/10.2214/AJR.08.1910 ArticlePubMed
• 9. Gefter WB, Post BA, Hatabu H. Commonly missed findings on chest radiographs: causes and consequences. Chest 2023;163:650-661. https://doi.org/10.1016/j.chest.2022.10.039 ArticlePubMed
• 10. Dotter CT. Teaching of chest radiology to medical students. Dis Chest 1956;29:691-692. https://doi.org/10.1378/chest.29.6.691 ArticlePubMed

Figure & Data

References

Citations

Citations to this article as recorded by