Medical imaging has numerous applications and plays an irreplaceable role in clinical diagnosis. The importance as well as the impact of neuroimaging in diagnosis and clinical research of Alzheimer's disease is still expanding. The prevalence of Alzheimer's disease has increased in the past decade. It is hence vital to weigh the existing tools according to the particular clinical complication. Presently used imaging tools with respect to Alzheimer's disease are structural MRI, SPECT, PET, fMRI and its variations. In this review, an overview of the techniques is provided, along with how the techniques compare with each other in terms of the data provided.
With the advent of public health awareness and various breakthrough in diagnostic methods, the life expectancy of humans in USA has increased from 49 years to over 76 years in the past century [1]. This has resulted in an increase in the proportion of people over the age of 60, which implies an increase in the susceptibility to Alzheimer' s Disease (AD), the most common form of dementia. Statistical research conducted by the Alzheimer's Association reveals that, as of year 2010, 5.6 million Americans are affected by AD; 5.1 million among them are aged over 65 while the rest possess early onset form of the disease. Though heart disease, cancer, HIV AIDS continue to remain the top ranked causes of deaths, they have all experienced a decline in the percentage of deaths they cause, over the years 2000 to 2006 while deaths due to Alzheimer's disease has increased by about 46%. For instance, number of deaths due to heart disease has dropped by about 11% in the year 2006 in comparison to year 2000 while AD accounts for 46% more deaths in 2006 when compared to year 2000. [2] (Figure 1)
Alzheimer's disease results in memory loss and cognitive impairment. The most striking features of AD with respect to brain anatomy are: a) Atrophy of medial temporal lobe, entorhinal cortex and hippocampus, b) widened ventricles, c) shrinkage of gyri (infoldings of the brain) and consequent widening of sulci (furrows in the brain) leading to volumetric changes in the affected brain.
The disease progresses by the atrophy of various regions of the brain; the functional loss in the patient parallels with the regions at which atrophy occurs.
Biochemical changes include a) neuritic senile plaques due to deposition of amyloid beta and b) formation of neurofibrillary tangles. Amyloid beta Figure 1: Trends in the number of deaths caused by each disease between years 2000-2006: From: 2010 Alzheimer's facts and figures by Alzheimer's Association
deposition occurs when the transmembrane protein on neurons, Amyloid Precursor Protein (APP) which is responsible for growth, survival and post injury repair of neurons, is cleaved by enzymatic activity to release peptides like amyloid- beta which are insoluble and form clumps near the neurons.[3]
Another biochemical complication associated with AD is the formation of neurofibrillary tangles. The microtubules contained within neurons are responsible for transport of molecules and maintenance of environment. These microtubules contain a specific protein Tau which is phosphorylated and helps in proper transport and stabilises the microtubules. Under pathological conditions of AD, Tau is hyperphosphorylated and hence forms tangles and disrupts the transport system by destabilising/ collapsing the microtubules. [3]
2. Diagnosis
Imaging of the brain and its parts (neuroimaging) is the ultimate most powerful diagnostic tool for AD. Patients demonstrating cognitive deterioration may, potentially be affected by other conditions like tumors, chronic subdural haematoma which are fatal, but reversible after treatment. Hence neuroimaging is invariably prescribed by physicians to screen patients according to treatment regimen to be followed. Neuroimaging also screens patients according to the type of dementia (vascular, frontoparietal etc)
Neuroimaging can be a) structural where imaging is done to identify the volume changes or b) functional where imaging is done to identify metabolic abnormalities like change in glucose metabolism, blood flow, deposition of substances, tangles etc. Both types of neuroimaging come handy in diagnosis and clinical research of AD. It is important to review and study the different tools that have been used and how they were modified to lead to the present day techniques. A comparison of the techniques used is mandatory to precisely state which is best suited to identify the disorder at its early stage, to distinguish AD from other similar disorders and to study progression of disease from MCI (Mild Cognitive Impairment) to AD and dementia. Mild Cognitive Impairment (MCI) is a condition in aged, in which cognitive abilities show deterioration. People affected by MCI possess an increased risk factor of developing AD in later stages. Presently, diagnosis is possible only after the disease symptoms are fully pronounced. By that time, considerable brain pathology and hippocampal loss would have already occurred[4] Hence it is crucial to develop methods to diagnose the disease in earlier stages. A presymptomatic diagnosis is a necessity to decrease the deaths due to AD. This can be made possible by using a positive biomarker for AD.
Figure 2: Representation of anatomical changes in the brain
2.1 Structural Neuroimaging of AD
Demarrs and colleagues have shown that amyloid- beta deposition is an early marker of in the pathogenesis of AD. [5] The event of neurodegeneration and subsequent atrophy of hippocampus and cortex follows the deposition of amyloid- beta plaques.[6,8] Structural neuroimaging is of great importance in the diagnosis of AD as the events of neurodegeneration, atrophy closely correlate to cognitive loss in subjects. The use of atrophy as a biomarker of AD is hence widespread.
Magnetic Resonance Imaging is the most commonly used method. Common structural analytical tools employed by sMRI are volumetry, voxel based morphometry, tensor based morphometry.[7]
2.1.1 Visual method:
The simplest analytical tool for scans is visual screening. Diagnosticians can compare the scans to detect atrophy in medial temporal lobe and hippocampus. But this method may not be effective in detecting quantitative differences in the volume.
2.1.2 Volumetry:
Manual tracing: Manual tracing of region of interest (ROI based volumetry), the anatomical boundaries of hippocampus can be done but they are very tedious and time consuming. The hippocampus, being most deteriorated during AD, its volume can be measured by carefully tracing the anatomical boundaries of hippocampus and using algorithms to perform high dimensional fluid transformations[7]. Commonly used software to measure volume is software suite Analyze 6.0 which can be programmed with a set of rules. These automatic and semi automatic techniques measure not only the thickness of the region of interest but also the global thickness. The disadvantage of this method is that only a single region of interest is used to consolidate the 3-dimensional information, which is spatially limited as it does not use the whole data available.
2.1.3 Voxel Based Morphometry (VBM):
The manual measurement of brain volume is tedious and time consuming. VBM is an automated method to measure/ compare volume of region of interest within a group. A T1 weighted MRI is performed and VBM uses statistical methods to compare volume at the voxel level in the scans. Once the MRI scans of different subjects are obtained, they have to be spatially normalised so that the statistical methods can be applied on them. Spatial normalisation is a process that ensures that all the scans obtained are matched spatially. This can be done by algorithms that transfer all scans obtained onto a template. Once this is done, the images are compartmentalised (gray, white matter, Cerebro Spinal Fluid CSF) and the compartment of interest (gray) is analysed at voxel level. This process is known as segmentation. The images are then subject to smoothing, wherein the intensity of each voxel is substituted by weighted average of surrounding intensities. Statistical analyses of the smoothed images are performed by methods like theory of Gaussian random fields and comparison is made. The main disadvantage of VBM is that the statistical method of comparison can be applied only on groups of subjects to provide information about atrophy. The method is inadequate and cannot be applied on individual subjects.[7,8,9]
2.1.4 Tensor based morphometry:
This method also uses statistical means to identify differences between various brain scans. A non linear registration algorithm aligns the scans obtained to a common space, which can be a standard high resolution brain atlas or a control's scan. Few other methods use a template that is generated from the average image from all control scans. The relative gain/ loss in brain volume is measured using statistical analysis.[10]
2.2 Functional neuroimaging
Functional neuroimaging is different from structural neuroimaging in that it measures parameters like metabolism of a tracer compound which is in turn related to the blood flow and perfusion in each region of brain.
2.2.1 SPECT:
Single Photon Emission Computed Tomography is a nuclear imaging technique in which the patient is given a dose of a radiopharmaceutical which emits gamma rays. The emitted gamma rays are recorded by an array of gamma cameras. The tracer used is meta stable 99Tc- HMPAO (Hexa methyl propylene amine oxime)
Regional Cerebral blood flow (rCBF) measurements based on retention of tracer distinguish between AD and other types of dementia. For instance, while frontotemporal dementia has abnormal perfusion in frontal and temporal region, dementia with Lewy bodies has occipital hypoperfusion, AD has abnormal perfusion in parietotemporal region. [11,12]
2.2.2 PET:
Positron Emission Tomography is a nuclear medicine technique that produces scans of organs by detection of gamma rays that are produced by the annihilation of positrons with electrons in the tissue. The technique is carried out by providing a dose of positron emitting radiopharmaceutical to the patient. When the annihilation of the emitted positron by the electrons takes place, 2 anti parallel gamma rays are produced. The fact that 2 gamma rays are produced improves the SNR and spatial resolution of image. Detection of these emitted gamma rays helps in tracing the distribution of the radiopharmaceutical in the organs. Commonly used radiopharmaceutical is Flurodeoxyglucose which contains radioactive isotopes 18F and 11C palmitate. FDG is a glucose analogue, which does not undergo further glycolysis due to the absence of hydroxyl group, which is substituted by the radioactive 18F. FDG is retained in the cells and shows up in the scans.
Apart from FDG, compounds like Pittsburgh compound B (PIB) can label the amyloid beta deposits in the AD affected brain specifically. 2-(1-{6-[(2-[F-18] fluoroethyl)(methyl)amino]-2-naphthyl}ethylidene)malononitrile (FDDNP) is another molecule that can be used in PET scans. FDDNP is claimed to label both the pathological hallmarks of AD, amyloid beta and neuro fibrillary tangles. [13,14]
2.2.3 fMRI:
As the glucose requirement varies with the brain region, the cerebral blood flow varies. The oxygen that is transported to the particular region also varies. It follows that the oxyhaemoglobin in the veins depend on the glucose requirement or activity of the brain. Since oxyhaemoglobin is diamagnetic and deoxyhaemoglobin is paramagnetic, each oxygen level in blood can be associated with a particular magnetic resonance signal. This technique in which Blood Oxygen Level Dependence (BOLD) is the biomarker to acquire images is a popular fMRI tool. Responses to various stimuli are measured by change in deoxyhaemoglobin. This enables mapping the particular region that experiences a change in rCBF.
Technique is performed by T2* weighted imaging with a stimulus according to the region to be mapped. One short coming in BOLD fMRI is that, there is a time lag between activation of a tissue and increased blood flow to that region. In fact, the oxygen concentration actually decreases in the activated regions and then shows an increase. This factor largely affects the temporal resolution of BOLD fMRI.
Perfusion fMRI aims at measuring rCBF by means of intravenous bolus tracking of a Gadolinium
Figure 3: PET images comparing temporal lobe uptake of [18F]FDDNP and FDG, a marker of glucose metabolism, in a patient with AD (Top) and a control subject (Botom). Note increased uptake and retention of [18F]FDDNP (arrowheads) in temporal lobes of the patient with AD, compared with those in control subject. The patient with
AD still demonstrates typical findings of decreased temporal (arrows) and parietal (not shown) FDG uptake.[14]
based contrast agent. Areas with the tracer show up with reduced T2* signals due to inhomogenity caused by its own magnetic field. Perfusion maps showing difference in blood flow under various conditions can be generated when perfusion MRI is first performed under control and then under activation stimuli. Caveats of perfusion fMRI are that it gives only the relative blood flow and flow comparisons and not absolute values, injection of magnetic bolus limits the repeatability of the procedure.
Arterial Spin labelling fMRI circumvents the above problem. In this technique, the water- based Hydrogen atoms present in the artery outside the slice of interest are magnetically labelled. T1 weighted scanning is performed. One limitation of this technique is its high slice acquisition time.
3. Comparison
SPECT involves exposing patients to radioactive materials, placing an upper limit to the dose. When compared to SPECT, the fact that 2 gamma rays are produced in PET improves the SNR and spatial
resolution of image[15]. PET aims at detecting amyloid beta deposits. It is known that amyloid
deposition is an early event in the pathology of AD. It follows that PET could be used to conduct early diagnosis and can also be used to study disease progression. Also, it is more expensive as a cyclotron has to be installed close to the place as the radionuclides have very short half life. PET involves use of radiotracers which places an upper limit on its repeatability. PET has a low temporal resolution and SNR when compared to fMRI[16]. Studies show that the longitudinal changes are much more pronounced in sMRI than in PET[7]. Also, studies published by Alzheimer's Disease Neuroimaging Initiative have shown that sMRI is more related to cognition than the imaging with CSF biomarkers.
4. Conclusion and Future perspectives
While some studies show an association between CSF biomarkers and sMRI methods[17], others donot. [18]This variation may be due to variability in procedure and methods. More studies need to be done in order to understand the reasons behind the mismatch. Ordering of the modalities and techniques depends on the specific situation and is highly subjective. Functional neuroimaging for AD with CSF biomarkers as well as structural neuroimaging have both proved to be of great diagnostic value. It is believed that both diagnosis performed together provides better discrimination of AD than either performed alone. Most helpful diagnostic measure is possibly a structural imaging combined with AD specific marker technique like PET PIB. Keeping in mind the high cost involved in the purchase of multiple equipments, a good solution to this problem could be a combined MRI- PET instrument that can simultaneously perform the scans. This is not a far-fetched idea; in fact, this has already been ventured by companies like Siemens. [19]
