Displaying perfusion MRI images as color intensity projections

Dynamic susceptibility-weighted contrast-enhanced (DSC) MRI or perfusion-MRI plays an important role in the non-invasive assessment of tumor vascularity. However, the large number of images provided by the method makes display and interpretation of the results challenging. Current practice is to display the perfusion information as relative cerebral blood volume maps (rCBV). Color intensity projections (CIPs) provides a simple, intuitive display of the perfusion-MRI data so that regional perfusion characteristics are intrinsically integrated into the anatomy structure the T2 images. The ease of use and quick calculation time of CIPs should allow it to be easily integrated into current analysis and interpretation pipelines.

💡 Research Summary

Dynamic susceptibility‑weighted contrast‑enhanced (DSC) MRI, commonly referred to as perfusion MRI, provides quantitative information on cerebral microvascular density and permeability, which is essential for assessing tumor vascularity, differentiating tumor progression from radiation‑induced necrosis, and guiding treatment decisions. However, the technique generates a large number of time‑resolved images (often hundreds), making conventional display methods—typically relative cerebral blood volume (rCBV) maps—cumbersome to interpret, especially because rCBV maps are usually presented separately from the anatomical T2‑weighted images that clinicians rely on for spatial context.

The authors propose the use of Color Intensity Projections (CIPs) as a novel visualization strategy for DSC‑MRI data. CIP is a technique originally developed for four‑dimensional CT and digital subtraction angiography (DSA) that condenses an entire temporal series into a single color image. For each pixel, the algorithm extracts the minimum, maximum, and mean signal intensity across the time series. The minimum‑maximum range determines the saturation (how vivid the color appears), while the mean value determines the hue: pixels whose intensity is close to the minimum are rendered red, those near the maximum appear blue, and intermediate values are mapped to yellow, green, or cyan. Because the mean signal intensity in DSC‑MRI is proportional to the relative CBV, the hue directly encodes perfusion magnitude, while saturation reflects the dynamic range of signal change. Consequently, red pixels denote rapidly filling vessels (low average signal), green/yellow indicate highly perfused tissue, and blue signifies low‑perfusion regions.

To demonstrate feasibility, the study examined a 52‑year‑old female patient who had undergone stereotactic radiosurgery for a solitary left frontal lobe metastasis. Six months post‑treatment, gadolinium‑enhanced T1‑weighted imaging showed new peripheral enhancement, raising the differential diagnosis of tumor progression versus radiation necrosis. A DSC‑MRI acquisition was performed using a T2 echo‑planar sequence, capturing images from the moment of contrast arrival up to 23 seconds later at 1‑second intervals. The entire CIP generation required less than two seconds of computational time, illustrating the method’s suitability for real‑time clinical workflows.

Both the CIP image and the conventional rCBV map identified relatively high perfusion on the lateral and dorsal aspects of the lesion, correlating with the contrast‑enhancing region on the T1‑weighted scan and supporting a diagnosis of tumor progression. However, the CIP offered a distinct advantage: the color‑coded perfusion information was overlaid directly onto the anatomical T2 background, allowing clinicians to assess perfusion heterogeneity in the context of brain structures without mentally registering separate images. This integration simplifies interpretation, reduces the risk of spatial misregistration, and may accelerate decision‑making in multidisciplinary tumor boards.

The discussion emphasizes three primary benefits of CIP: (1) speed—only three simple statistics per pixel are required, avoiding complex deconvolution or model‑based calculations; (2) intuitiveness—clinicians can instantly gauge both the magnitude and variability of perfusion from a single color image; and (3) anatomical integration—by embedding perfusion data within the native T2 anatomy, CIP eliminates the need for separate overlay tools. The authors also acknowledge limitations. CIP provides only relative, not absolute, perfusion values; therefore, inter‑patient or longitudinal comparisons require additional calibration. Low‑variance regions appear gray, potentially obscuring subtle pathology. Color perception issues (e.g., color‑blindness) and the choice of hue‑saturation mapping may affect reproducibility across institutions. Moreover, the robustness of the hue‑to‑CBV relationship across different MRI protocols (varying TR, TE, flip angle) remains to be validated.

Future work outlined includes (a) a larger, multi‑center qualitative and quantitative comparison of CIP versus standard rCBV, K^trans, and other perfusion metrics; (b) development of automated color‑balance algorithms to standardize hue mapping across scanners; (c) integration of CIP generation into PACS or treatment‑planning software for seamless real‑time use; and (d) assessment of CIP’s utility in other clinical scenarios such as stroke, epilepsy, and neurodegenerative disease where perfusion dynamics are relevant.

In conclusion, the study demonstrates that Color Intensity Projections can condense the rich temporal information of DSC‑MRI into an easily interpretable, anatomically integrated color image. With sub‑second processing times and intuitive visual encoding of both perfusion magnitude and variability, CIP holds promise as a practical adjunct to conventional rCBV maps, potentially improving diagnostic confidence and workflow efficiency in neuro‑oncology and beyond. The authors have filed a patent on the method and are pursuing further validation to support broader clinical adoption.