Low-Dose CT Offers MRI-Free Tau PET Quantification

Low-dose computed tomography could provide a reliable anatomical reference for tau positron emission tomography when magnetic resonance imaging is unavailable or unsuitable. A multicentre comparison published in The Lancet Medical Imaging and Theranostics assessed a CT-based processing method against standard MRI-based approaches. The analysis included 387 participants from six sites, ranging from cognitively unimpaired adults to people with mild cognitive impairment or dementia. Across two tau tracers, CT-based measurements closely matched MRI-based results. The method also produced similar tau-positivity classifications and preserved links between tau levels, amyloid burden and memory performance. 

Using CT Instead of a Separate MRI 

Quantitative tau positron emission tomography usually depends on structural magnetic resonance imaging. MRI helps define reference regions, align images and divide the brain into anatomical areas. However, an additional MRI scan increases scheduling demands, costs and the need for patient cooperation. MRI may also be unsuitable for people with certain implanted devices or severe claustrophobia. It may also be missing from older imaging datasets. 

Must Read: Wearable PET System Advances Mobile Brain Imaging 

The alternative method uses the low-dose CT scan already acquired during a PET–CT examination. This CT scan is normally used for attenuation correction. The image is processed to make the boundary between the brain and skull clearer, then aligned with a standard anatomical template. The same transformation is applied to the tau PET image. This allows uptake measurements to be calculated without a structural MRI. 

Participants underwent PET–CT with the tau tracers MK6240 and flortaucipir, as well as MRI and amyloid PET. The CT-based method was compared with two MRI-based workflows. One used template-based processing and the other used FreeSurfer segmentation in each participant’s own anatomical space. Comparisons covered three Alzheimer’s disease-related composite regions and 34 individual brain regions. They assessed measurement agreement, equivalence and consistency in classifying scans as tau-positive or tau-negative. The comparison also examined whether biological relationships remained stable across the different workflows. 

Measurements Closely Match MRI-Based Results 

CT-based uptake measurements showed very strong agreement with the template-based MRI workflow across the medial temporal, neo-temporal and later neocortical regions. Average brain maps also showed similar patterns of tau deposition across participants with different levels of cognitive impairment. Agreement remained strong when the CT method was compared with the more demanding FreeSurfer workflow. Formal testing confirmed equivalence across all three composite regions for both tracers. 

The regional analysis gave similar results. Across 34 brain regions, agreement remained high for MK6240 and flortaucipir, with differences that were generally negligible or small. Early Alzheimer’s disease regions, including the transentorhinal and entorhinal cortices, also showed high agreement. Some peripheral neocortical regions showed larger differences, but these were still small. A limited proportional bias appeared in the medial temporal region at higher uptake levels. 

Tau-positivity classifications agreed in 92% to 97% of participants, depending on the tracer, region and MRI comparison. Most disagreements occurred close to the classification threshold rather than showing a consistent bias. Similar findings appeared when another method was used to define positivity thresholds. The CT-based and MRI-based workflows also preserved similar associations between tau uptake, amyloid burden and memory performance. A small number of CT-based cases could not be processed, while both MRI-based workflows completed all cases. 

Potential Uses and Important Limitations 

CT-only processing could extend quantitative tau PET to people who cannot undergo MRI. This includes some older adults with cardiac devices, neurostimulators or other MRI contraindications. It could also support analysis of older tau PET datasets in which MRI was not acquired because of technical, scheduling or logistical constraints. Because low-dose CT is already part of routine PET–CT imaging, no extra scan is needed solely to align the PET image. 

The method could reduce imaging and processing demands when the main purpose is tau PET quantification rather than detailed structural assessment. It also avoids the time and resources needed for individual MRI segmentation while producing similar measurements. Retaining low-dose CT images in clinical archives could therefore support later analysis of PET examinations. 

MRI remains the preferred anatomical reference when it is available. It provides information on regional atrophy, cortical thickness and brain volume that low-dose CT cannot provide. CT-based processing also cannot perform partial-volume correction and may underestimate tau burden in people with severe atrophy. Accurate alignment between PET and CT remains essential, and poor alignment must be corrected before processing. Advanced cortical atrophy or enlarged ventricles may also make CT-to-template alignment less reliable. The cross-sectional design means that the method still needs testing for its ability to measure change over time. 

Low-dose CT can support tau PET quantification with results that closely match standard MRI-based processing across two tracers and many brain regions. It also provides similar tau-positivity classifications and preserves associations with amyloid burden and memory. The main benefit is the ability to quantify tau PET when MRI is unavailable, unsuitable or missing from existing datasets. MRI remains preferable when detailed structural information is required. Further work is needed to confirm whether CT-based processing can measure changes in tau levels reliably over time. 

Source: The Lancet Medical Imaging & Theranostics 

Image Credit: iStock