BAAD (Brain Anatomical Analysis using Diffeomorphic deformation) is based on a VBM (voxel-based morphometry) algorithm to extract features of the brain shape. It runs on Windows OS and requires no special preparation. Download and use it freely. It comes standard with AAL, Brodmann, and LPBA40 regions of interest (ROIs), as well as independently created regions of interest for white matter and white matter lesions (deep white matter and periventricular white matter). Users can also create their own ROIs and register them in BAAD. In addition to statistical analysis, z-values and volumes (ml) for each ROI are calculated.
BAAD supports the clinical diagnosis of Alzheimer's disease (AD) from brain MRI. The BAAD-AI, which have been trained using the North American ADNI database, outperformed radiologists for AD diagnosis in a structural brain MRI review.
BAAD server can be installed in hospitals. This enables automatic BAAD analysis and reduces the daily workload of the hospital. Medical doctors can check the results of the BAAD along with the ordered MRI images via PACS. Furthermore, when the BAAD viewer is installed on a PC using as an electronic medical record, various analyses can be performed on the spot. For example, inputting the MMSE score will help the clinical diagnosis or doctors can change the brain section and threshold at will. Data is managed and stored on a case-by-case basis, and the data are passed to the BAAD-AI that deals with time series, helping to predict disease onset.
BAAD is also useful for research. Through BAAD, SPM12 and CAT12 can be used freely without installing Matlab. You can convert multiple series of DICOM data into NIfTI files at once, or edit DICOM files while viewing the images. If FLAIR images of the same case exist, they can be automatically paired with 3DT1-weighted images during segmentation for highly accurate segmentation, and white matter lesions can be measured by region.

Alzheimer's disease (AD) can be conceptualized as a biological and clinical continuum from the preclinical phase (clinically asymptomatic subjects with AD pathology) to the clinical phase. Therefore, we trained the BAAD-AI by labeling AD and progressive MCI as being in the AD spectrum, and cognitively normal and stable MCI as being in the non-AD spectrum. The AI processes hundreds of information related to the structure of the brain, and presents a possibility of AD by a value from 0 to 1as Alzheimer's score (ADS). An ADS > 0.5 suggests the possibility of AD spectrum.
BAAD-AI outperformed radiologists for AD diagnosis in a structural magnetic resonance imaging review. (See references)

In schizophrenia, atrophy of the frontal lobes and insular gyrus appears. AI allows us to detect these features earlier. By using AI, it will be possible to capture these characteristics earlier. As shown in the VBM result, globus pallidus may hypertrophied in some cases.

There are several examinations to evaluate aging of organs, such as advanced glycation end products (AGEs) from blood, vascular age by CAVI/ABI, osteoporosis by x-ray, and lung age by spirometer. By the way, how should "brain age" be assessed? Brain function or cognitive ability affects the quality of daily work and life. The human brain, weighing approximately 1.4 kg, is an amazing feat of engineering with approximately 100 billion neurons interconnected via trillions of synapses. During the first few years of life, the brain forms more than 1 million new neural connections every second. With aging, neurons and synapses decrease, making it more difficult to learn new things and plan or accomplish multiple tasks simultaneously. Decline in experiential and working memory, processing of external information and judgment, and nerve-derived motor functions occurs.
To evaluate brain aging, tests for cognitive function and physical ability are considered, but a vast number of tests and complex correction algorithms are required because differences in educational and environmental backgrounds, and arthritic disease and muscle weakness may also affect motor function. If we think the brain that forms synapses as computer hardware, its ability can be evaluated using software that draws out the limits. This can be done with a computer, but it can be painful to humans to tasks that are close to their limits. On the other hand, measuring brain volume as a hardware aspect is relatively easy. Brain volume is known to decrease with age, at a rate of about 5% per decade after the age of 40, with a particularly accelerated decline after the age of 70. Although brain size does not necessarily correspond to its ability, considering that the number of neurons and their networks are reflected in brain volume, brain size could be an indicator of not only the capacity normally exhibited but also its potential.
BAAD has successfully assessed the biological brain age in individuals by training AI on morphometrical changes of brain development and aging from brain MRI of more than 7,000 healthy subjects ranging in age from 5 to 98 years.

Brain atrophy is associated with the risk of stroke. In subjects with high stroke risk factors, reductions in frontal and posterior brain volumes as well as hippocampal volumes have been observed {24582641}, {22984010}, {21810696}, {28498826}, {26239040}, {21035552}. In UK Biobank (n = 9722, age range 44-79 years), a report examining the association of multiple vascular risk factors (VRF; smoking, hypertension, pulse pressure, diabetes, hypercholesterolemia, obesity, waist/hip ratio) with brain structure and diffusion MRI markers found that the number of VRF was additively associated with reduced gray matter volume and worse in white matter conditions. Higher numbers of VFRs were associated with decreased volume of frontal and temporal lobe gray matter, decreased subcortical volume, increased white matter lesions, and worsened white matter microstructure in association and thalamic pathways. Years of smoking, hypertension, and diabetes have shown particularly strong associations {30854560}. The Framingham Stroke Risk Profile (FSRP), a stroke risk prediction tool, is an index of vascular risk factors that includes older age, gender, high systolic blood pressure, antihypertensive treatment, left ventricular hypertrophy on ECG, presence of cardiovascular disease, current smoking, atrial fibrillation, and diabetes. In both cross-sectional and longitudinal analyses, higher risk factors by FSRP were associated with brain atrophy in each age group {30232248}, and the difference between MRI-predicted age and actual age (BAG) was significantly associated with systolic blood pressure, smoking, and C-reactive protein (CRP), which are related to brain aging. It has been shown {34626047}. There have also been reports of an association between obesity and diabetes and brain atrophy; type 2 diabetes was associated with decreased speed of thought processing, white matter lesions expansion, and brain atrophy {34588209}. In an epidemiological study from 45 independent centers, cross-sectional analysis showed pooled weighted mean differences in total brain volume and gray matter volume for obese/overweight participants were -11.59 and -10.98, respectively. Total brain volume was negatively correlated with body mass index (BMI) and abdominal circumference, and hippocampal volume has been negatively correlated with BMI, abdominal circumference, and waist-to-hip ratio {34391946}. Decreased whole brain volume is an independent risk factor for death, independent of cerebrovascular disease and multiple risk factors. Brain volume is thought to be an integrative indicator reflecting various health conditions {2535359930}. Across all racial and ethnic groups, older age, smoking, hypertension, and diabetes were strongly associated with white matter lesions {35352569}. Lower glomerular filtration rate (eGFR) was associated with decreased cortical volume and increased white matter lesions {36179945}. In asymptomatic middle-aged and older adults, progressive brain atrophy 20 years before dementia onset has been shown to be associated with increased risk of future dementia {31806724}. 

Aging score (-----) indicates whether you are older (+) or younger (-) compared to your actual age. For example, a dotted line of +8.9 indicates that your brain has increased by 8.9 years (aging).
The four diamond-plots show the relative age of the frontal lobe, hippocampus, insular gyrus, and white matter lesions, from left to right. Higher values (+) indicate worse. Volume doesn't necessarily reflect brain capacity, but it does suggest a relative change with your innate capacity.
◆ The frontal lobe is involved in judgment, thinking, creativity, social behavior, and mental arithmetic.
◆ The hippocampus is related to your episodic memory and short-term memory. It declines with age , especially in Alzheimer's disease.
◆ The insular cortex is related to primitive perception and emotion. It becomes more difficult to control with aging (+).
◆ White matter lesions reflect the aging of the small blood vessels . It increases rapidly after age 50. Some of them may include small infarcts. The higher number reflects the aging of the small blood vessels.

Percent volume change/year：
Annual rate of your brain volume comparing to the previous MRI. Brain atrophy is usually about 0.5%, but varies with age. Please refer “your brain volume” graph.

MR images were preprocessed by VBM to standardize and simplify the features. Specifically, this included ensuring a unified brain shape, converting the volume into signal intensity, and adjusting the brain volume according to total intracranial volume (TIV) and age.
Structural parcellations such as the AAL, LPBA, and Brodmann area were adopted. These atlases were expected to reflect the unit of the neuronal network to some extent. Using multiple atlases facilitates optimal weight distribution of features during machine learning, supposedly compensating for the discrepancy between anatomical and pathology-specific boundaries.

We recommend installing BAAD server system in your clinic. With BAAD server, you can manage all BAAD-related data. The BAAD Client monitors the BAAD server, and when it detects new DICOM files, it automatically starts analysis and sends the results back to PACS via BAAD server. This automated system can also be adapted to other applications such as VSRAD. Therefore, physicians can check the results of BAAD and/or VSRAD together with MR images on their electric medical record through PACS.
With the BAAD viewer, further analysis can be performed on a PC equipped with an electronic medical record. In addition, by inputting MMSE score into the BAAD viewer, BAAD-AI will display advanced results based on the brain MRI results. The results of the analysis are stored and managed on the BAAD server, allowing for long-term analysis.

You can use SPM12 or CAT12 from BAAD. Permutation test and TFCE are available through these programs.
In addition to VBM, surface-based morphometry (SBM) and tensor-based morphometry (TBM) can be performed.
Measurement of the volumes of white matter lesions is available, they can be measured separately as periventricular and deep white matter lesions. For hydrocephalus, the volume of the lateral ventricles can also be measured.
The biological age of the brain can be estimated by applying machine learning to magnetic resonance imaging (MRI) data. BAAD will show a patient’s biological age using eXtreme Gradient Boosting (XGBoost). The gap between biological and chronological brain age may reflect past and ongoing neurobiological aging processes.

BAAD is easy to get started. After downloading, follow the instructions to install it. Double-click the BAAD icon to start, then a top panel will open. Please input the data to analyze, and click the “analyze” button. Then it runs fully automatically.

It is useful to set up a DICOM receiver that receives DICOM files directly from the MRI console, and it is even more useful to introduce our BAAD server system. If you want to build your own DICOM receiver, you may use a free software such as Conquest DICOM software. It is also possible to read DICOM files from CDs. In some cases, it may be hard to import DICOM files from CD, because they may have been specially processed by each vendor. In such a case, try using BAAD's DICOM import function.

Since BAAD uses the IXI database, it is ideal to image under the same conditions close to the IXI dataset.
As the voxel size is 0.9375 x 0.9375 x 1.2 mm, it is better to take an image with a size close to 1 mm cubic. The T1-weighted three-dimensional MP-RAGE sequences is the standard and is often used: For example,
1.5T (Gyroscan): TR/TE=9.813/4.603, FA=8, FOV=240, 256x256x150
3.0T (Intera 3T): TR/TE=9.600/4.603, FA=8, FOV=240, 256×256×150
For 3D brain imaging, it is necessary to keep the high contrast between gray matter and white matter. Sagittal section is recommended, but horizontal and coronal sections can also be analyzed.
BAAD uses the Markov-MAP method for segmentation to accurately capture pathological changes. If FLAIR images are available (as thin slices as possible, gapless is better), they can be complemented with 3DT1 images to extract and correct white matter lesions. By the way, BAAD measures the volume of white matter lesions.

To manage the results of BAAD on a case-by-case basis, we have a BAAD server system that can work with BAAD clients and BAAD viewers to handle electronic medical records. You can build the system by yourself or get support from us. Please feel free to contact us.

The panel below shows the image quality (IQR) results of the IXI database used as a control group for BAAD. The median is 85.59 and the range is from 86.06 to 80.76.
The next panel shows the IQR calculated as a weighted average from noise level, INU level, and image resolution. For more information, please refer to the CAT12 website. IQR gives you an idea of the quality of the MR image. An IQR of 80 or higher is considered good, if possible.

Modified from http://www.neuro.uni-jena.de/cat12-html/cat_methods_QA.html

Textbook for VBM

[Theory]
Bayes' theorem, maximum likelihood estimation, mutual daughter cells, Markov probability fields, linear models, statistics in SPM (multiplicity of tests, correction for multiple tests, test levels, permutation tests, TFCE), coordinate transformation
[Basics]
File format of image data, coordinate space of brain, preprocessing for analysis (AC-PC correction, correction of signal value inhomogeneity, tissue segmentation, Gaussian mixture model, warping to MNI space, difference between modulation and concentration, intracranial volume, volume calculation of cerebral white matter lesion) (volume calculation of cerebral white matter lesions), VBM, TBM, SBM, population intelligence
[Practice]
How to install and use BAAD (including how to manage data, how to create custom ROIs, etc.)
How to use SPM12 (VBM and TBM with SPM12, geodesic shoot, Dartel, conversion to MNI space, etc.)
SBM methods (VBM with CAT12, SBM methods, etc.)