Advancing Alzheimer’s Diagnosis: How AI is Enhancing Brain Imaging from Routine CT scans

Artificial Intelligence (AI) is a type of computer technology that learns to recognise patterns in huge amounts of data. It is a versatile tool that enhances a wide range of tasks by automating complex processes, analysing vast amounts of data, and making intelligent decisions. For example, in healthcare, AI helps diagnose diseases by analyzing medical images like X-rays and magnetic resonance images (MRIs) with high accuracy; in finance, it detects fraudulent transactions by recognizing unusual patterns in real-time; in customer service, AI-powered chatbots handle routine inquiries instantly, freeing up human agents for complex issues; in creative fields, AI generates music, artwork, or even writes text based on user input; and in everyday life, virtual assistants like Siri or Alexa help manage schedules, answer questions, and control smart home devices.  

These examples demonstrate how AI can be applied broadly to improve efficiency and innovation across diverse areas. AI systems, particularly those based on deep learning (a type of AI that mimics how the human brain processes information), analyse complex brain imaging data from modalities like MRI and positron emission tomography (PET) scans. These algorithms identify early signs, such as atrophy (shrinking) in specific brain regions or the accumulation of toxic proteins that are indicative of Alzheimer’s disease and related dementias. Moreover, AI can assist in differentiating between various types of dementia by recognising distinct imaging patterns, thus supporting more tailored treatment strategies. 

Beyond diagnosis, AI also plays a critical role in predicting disease progression by integrating longitudinal imaging data (images taken over several months or years) with clinical assessments. This predictive capability empowers healthcare providers to design personalised care plans and optimise treatments. The adoption of AI in dementia diagnosis not only enhances diagnostic precision but also accelerates the assessment process, allowing clinicians to manage larger patient volumes more efficiently. Ultimately, these technological advances hold the promise of improving patient outcomes by facilitating earlier intervention and more informed clinical decision-making. 

The use of AI to improve brain tissue analysis on computer tomography (CT) scans makes them more comparable to magnetic resonance imaging (MRI) scans, which are currently the gold standard in the diagnostics of neurodegenerative diseases. 

Why Brain Imaging Matters in Alzheimer’s Disease? 

Magnetic Resonance Imaging (MRI) is a standard tool for assessing Alzheimer’s because it provides high-resolution (very detailed) images of brain structures, allowing clinicians to detect hallmark signs such as: 

• Atrophy of gray matter, especially in the hippocampus and cortex (areas of the brain important for memory and thinking)
• Changes in white matter connectivity (important for how different brain regions communicate)
• Alterations in cerebrospinal fluid spaces (fluid-filled areas around the brain and spinal cord)

However, MRI is not always available or suitable. Some older adults may be unable to undergo an MRI due to medical implants, claustrophobia, or frailty. In contrast, computer tomography (CT) is fast, widely accessible, and commonly used in emergency or routine care. 

Yet they traditionally lack the detail needed for precise dementia diagnosis. 

Clinical Applications: Bridging the Gap Between CT and MRI with AI 
A recent study by our team tackled this issue by using AI to “translate” CT scans into more detailed brain images. Researchers trained two advanced models – nnUNet and MedNeXt – to learn from thousands of paired CT and MRI scans, including those from the Gothenburg H70 Study and Singapore’s Memory Clinic Cohort. These models were taught to predict brain structures normally visible on MRI using only CT images. 

The results were promising. MedNeXt, the more powerful model, could accurately identify key brain regions tissues like grey matter, white matter, and cerebrospinal fluid, even in people with dementia. This suggests that CT scans, when enhanced with AI, could become a more powerful tool for detecting brain changes associated with dementia – making early diagnosis more accessible worldwide. 

Going Even Further: AI Maps the Chemistry of Alzheimer’s Plaques 

In a separate breakthrough study lead by Jörg Hanrieder and Michael Schöll from the University of Gothenburg,  AI was used to study amyloid plaques, the sticky protein clumps that build up in the brains of people with Alzheimer’s. While amyloid has long been known as a key protein linked to the disease, it remained unclear why some individuals with plaques develop symptoms, while others do not. 

Using a combination of high-resolution microscopy (powerful imaging tools that show tiny details) and mass spectrometry imaging (a method that shows the chemical makeup of tissues), the team examined individual plaques in donated brain tissue. These advanced techniques allowed them to analyse both the shape and chemical makeup of each plaque. AI then classified thousands of plaques based on subtle differences in their appearance and protein content. 

What they found was striking: a specific type of plaque, known as coarse-grained plaques – appeared only present in people who had symptoms of AD. These plaques contained high levels of certain damaging proteins (such as Aβ1-40 and Aβ3pE-40) and showed clear signs of disrupted nerve cells, which are closely linked to memory loss and other symptoms of dementia. 

This discovery opens a new chapter in Alzheimer’s research, allowing scientists to not only see amyloid plaques, but also understand which ones may be truly harmful. In the future, this could help us better predict who is at risk, guide the development of more targeted treatments, and improve the way we diagnose the disease. 

Towards a More Insightful Future 

Together, these advances show how AI is beginning to transform Alzheimer’s diagnosis and research, from improving routine brain scans to uncovering the molecular details of disease.  

While we are not yet at the stage of fully relying on AI for clinical decisions, it is proving to be a valuable tool in supporting healthcare professionals. 

As the global population ages, AI-powered tools can help clinicians extract more meaningful insights from widely available data, broadening diagnostic reach and supporting more personalised and effective care for those at risk of the disease.