Deconv_from_aif brick¶
Deconvolution of the tissue response curve with AIF¶
MRI perfusion imaging by deconvolution using an Arterial Input Function (AIF) is a method used to evaluate blood flow and perfusion in tissues, often in the brain, to diagnose and assess conditions like stroke, tumors, or other vascular abnormalities. This method is part of a broader category of dynamic susceptibility contrast (DSC) MRI techniques.
DSC-MRI involves the injection of a contrast agent (usually gadolinium-based) and the rapid acquisition of MRI images as the contrast passes through the blood vessels. This generates time-series data that reflect how the contrast agent is distributed in the tissue over time.
AIF represents the concentration of contrast agent over time in a feeding artery. It is used as a reference to determine the blood supply to the tissue.
Deconvolution is used to separate the tissue response from the AIF, thereby allowing the calculation of perfusion parameters like cerebral blood flow (CBF), cerebral blood volume (CBV), mean transit time (MTT), time to peak (TTP), time to maximum (Tmax) and bolus arrival time (T0).
Inputs parameters:
- func_file
T2* functional Magnetic Resonance Imaging (fMRI) experiment recorded during gadolinium bolus. Ideally, the data will have been pre-processed (realignment, segmentation, etc.). An existing, uncompressed file (valid extensions: [.img, .nii, .hdr]).
ex. '/home/username/data/raw_data/swrfunc.nii'
- aif_file
The AIF (a file in .json format) typically obtained previously with the Make_AIF brick.
ex. '/home/username/data/raw_data/swrfunc_aif.json'
- mask_file
The mask (at the resolution of func_file) used for the perfusion deconvolution (valid extensions: .nii).
ex. '/home/username/data/raw_data/mask_swc1_anat_003.nii'
- perf_normalisation
If perf_normalisation is not a number, no CBV normalisation is performed. Otherwise, the value of perf_normalisation will be used to normalise the CBV (and CBF) maps (any value is accepted).
ex. 5
- zero_pad
Zero padding factor (deconvolution parameters), a factor by which the original data is extended with zeros (an integer). A high zero-padding factor increases the signal length, potentially improving the precision of the deconvolution but also increasing computational complexity. A low factor (like 1, or no zero-padding) may result in less accurate outcomes, with lower frequency resolution and greater sensitivity to edge artifacts. The default zero_pad_fact is set to 2, meaning the signal length is doubled, which strikes a balance between precision and computational performance.
ex. 2
- oscil_th
Oscillation index threshold (deconvolution parameters), a threshold value that determines the level of acceptable oscillation during the deconvolution process (a float). During the OSVD (Oscillatory Singular Value Decomposition) deconvolution, the algorithm calculates the second derivative of the deconvolved signal (which highlights oscillations). The oscil_th sets a threshold for how much oscillation is acceptable in the signal. If the level of oscillation exceeds this threshold, the algorithm continues adjusting the truncation of singular values to reduce the instability. A low oscil_th will result in stricter filtering, meaning that more singular values (associated with oscillations) will be truncated. This can reduce noise but may also lose some of the true signal. A high oscil_th allows more oscillation, retaining more singular values and potentially preserving more of the original signal. However, this increases the risk of noise or instability in the output. Don’t change the default value (0.1) if you don’t know what you’re doing!
ex. 0.1
- bat_window_size
Number of time points (or dynamics) used to determine the bolus arrival time (t0) in the MRI signal analysis (an integer). Acts as a sliding window that moves in the time dimension of the MRI data, allowing the algorithm to analyse the local temporal behaviour at each voxel. A large window provides smoother estimates of the signal, but can delay the detection of sudden changes, such as the arrival of the bolus. It increases stability, but can also “blur” the sharp transition at bolus arrival. A smaller window size will be more sensitive to rapid changes, but may also pick up more noise or fluctuations in the signal. The default value (8) is a moderately sized window to smooth out short-term fluctuations while detecting significant changes in the MRI signal value.
ex. 8
- bat_th
Threshold multiplier (a float) used to detect significant changes in the signal intensity of MRI data (controls the sensitivity of the bolus arrival detection). A high bat_th value (e.g., bat_th == 3.0) would make the algorithm less sensitive, meaning it will only consider larger, more pronounced signal drops as valid bolus arrival points. This reduces false positives but may miss subtle or gradual bolus arrivals. A low bat_th value (e.g., bat_th == 1.0) would make the algorithm more sensitive, catching even smaller signal drops. However, this may lead to false positives, where noise or natural fluctuations in the signal are incorrectly identified as bolus arrival. The default (2.0) value ensures only signal drops larger than 2 standard deviations below the mean are considered significant, providing a balance between sensitivity and robustness against noise.
ex. 2.0
Outputs parameters:
- CBV_image
Cerebral Blood Volume (a file with .nii format) is a parameter that measures the total volume of blood present within a given volume of brain tissue, in milliliters of blood per 100 grams of brain tissue (mL/100g). CBV provides important information about the vascularity and perfusion characteristics of brain tissue.
ex. '/home/username/data/derived_data/swrfunc_CBV_deconv.nii'
- CBF_image
Cerebral Blood Flow (a file with .nii format) is a parameter that measures the rate at which blood is delivered to the brain tissue, in milliliters of blood per 100 grams of brain tissue per minute (mL/100g/min). CBF provides valuable information about the brain’s blood supply and is used to assess the adequacy of cerebral perfusion in various neurological conditions.
ex. '/home/username/data/derived_data/swrfunc_CBF_deconv.nii'
- MTT_image
Mean Transit Time (a file with .nii format) represents the average time (s) it takes for blood to pass through a given region of tissue. It is an important indicator of the efficiency of blood flow and is used alongside other perfusion metrics like CBF and CBV to assess the health of brain tissue.
ex. '/home/username/data/derived_data/swrfunc_MTT_deconv.nii'
- TTP_image
Time to Peak (a file with .nii format) reflects the time (s) it takes for the contrast agent to reach its maximum concentration in a given voxel after its arrival. This metric provides insight into the dynamics of blood flow and is particularly useful in evaluating conditions like stroke, tumors, and other cerebrovascular disorders.
ex. '/home/username/data/derived_data/swrfunc_TTP_deconv.nii'
- Tmax_image
The Time to Maximum (a file with .nii format) represents the time delay (s) between the arrival of contrast agent in the arterial input function (AIF) and the peak of the residue function (the tissue response after deconvolution).
ex. '/home/username/data/derived_data/swrfunc_Tmax_deconv.nii'
- T0_image
The Bolus Arrival Time (a file with .nii format) represents the time (s) at which the contrast agent first arrives at a particular voxel in the tissue. It essentially marks the onset of contrast passage through each voxel.
ex. '/home/username/data/derived_data/swrfunc_T0_deconv.nii'