HEEH-V1: an open-source DICOM research viewer with reproducible quantitative imaging workflows: technical validation

In the world of medical research, scientists rely on digital images to understand the human body without making an incision. These images, known as DICOM files, are the standard language that hospitals use to share scans like CTs and MRIs. However, when researchers want to measure specific details within these images—such as the density of a tissue or the texture of a tumor—they must use software to perform the calculations. A major challenge in this field is ensuring that different software programs produce the exact same numbers when analyzing the same picture. If one program says a tissue is slightly denser than another, it can lead to conflicting research results. To solve this, scientists often use "phantoms," which are objects designed to mimic human tissue but contain no actual patients, allowing them to test if their tools are consistent and reproducible.

A researcher named Mohamad Yousef Heeh has developed a new, free software tool called HEEH-V1 to help with this work. This application is designed for scientists and students to view medical images, take measurements, and run quantitative analyses. The core goal of the new study was not to prove that this software is better than existing tools or ready to diagnose patients, but simply to verify that it works correctly and consistently under specific conditions. The author wanted to see if the numbers produced by HEEH-V1 matched those from other established, trusted programs when looking at the same data. To do this, the team ran a series of rigorous checks using a public collection of CT scan images of a phantom, which is a specialized object used to test imaging equipment.

The researchers put the software through a comprehensive battery of tests to see if it could handle the complex task of reading and analyzing medical images. First, they ran the software against a set of 335 automated checks designed to ensure the code itself was functioning without errors. The software passed every single one of these tests without skipping a beat. Next, the team compared the results of HEEH-V1 against a well-known program called PyRadiomics. They took a specific set of 172 slices from a CT scan of a phantom and asked both programs to calculate four basic statistical values for a specific area on each slice: the average value, the standard deviation, the lowest value, and the highest value. These are fundamental measurements that describe the distribution of pixel values in an image.

The comparison revealed a perfect match. Across all 172 slices and all four types of measurements, the numbers generated by the new software were identical to those from the established program. The difference between the two sets of numbers was so small it was effectively zero, well within the tiny margin of error allowed for such precise scientific work. The researchers also tested the software's ability to handle specific reference standards used in the field of radiomics, known as IBSI configurations. In these tests, the software successfully processed 270 specific reference rows, matching the expected outcomes perfectly. While some other rows could not be compared because the features did not have a direct match, the ones that were tested all passed.

Beyond just matching numbers, the study also checked if the software could correctly identify the air surrounding the phantom, which should have a specific density value. The software measured the air on every slice and found values that were very close to the expected standard, with the largest difference being only 14.39 units, well within the acceptable range set for the study. The team also verified that the software could safely export the data without corruption and could detect when someone tried to tamper with the digital files. In these security checks, the software correctly accepted all valid files and rejected every single one of the 11 intentionally corrupted test files.

Despite these successful tests, the author is very clear about what this study does not prove. The results confirm that the software works reproducibly for the specific tasks and data tested, but they do not show that the tool is ready for use in a hospital to diagnose patients or plan treatments. The study did not test the software on real patients, nor did it compare its performance to every other imaging platform on the market. The findings are limited to the specific phantom images and the four statistical features that were measured. The author explicitly states that this work does not demonstrate clinical superiority, full compliance with every possible standard, or general privacy effectiveness beyond the specific tests performed.

The ultimate value of this work lies in its transparency and its contribution to the foundation of reproducible science. By providing a free, open-source tool and proving that it can generate consistent, verifiable results against known standards, the researcher offers the scientific community a reliable way to inspect and validate their own imaging workflows. The study serves as a technical validation, showing that the software behaves as expected under controlled conditions. It establishes a baseline of trust for the specific paths tested, allowing other researchers to build upon this work with the confidence that the underlying calculations are sound, while leaving the broader questions of clinical application and widespread adoption for future investigation.