Pre-process and analysis tab

This tab is part of the T2 Relaxation analysis settings dialog.

Noise reduction

Apply baseline correction:

Include an extra model parameter to account for an offset in the minimum (larger than zero) image intensity introduced by image noise

Eliminate outliers (spikes):

The highest 2% of the voxel intensities of the output images are set to the intensity of the 98-percentile value.

Apply spatial smoothing:

Nearest neighbour spatial smoothing is applied to the input data prior to analysis

Apply temporal smoothing:

Time-domain smoothing is applied to input data prior to analysis. The degree of temporal smoothing can be adjusted by the slider

Noise level

Apply noise level cutoff

Voxel intensities below the specified cutoff value are excluded from the analysis

Auto-detect noise level:

The noise level in the input data is automatically estimated using the Otsu’s method.

Noise level:

Override the auto-detected noise level by adjusting the slider

Eliminate non-connected voxels:

Performs seed growing from, starting from the image volume centre of gravity and eliminating all voxels not connected to the centroid voxel

Least squared weighting

Determines the relative weighting in the iterative least squares curve fitting procedure of the individual datapoints of the input data

Equal weighting for all datapoints: all datapoints are weighted equally, independent of amplitude

Increasing weighting of higher amplitudes: higher amplitude values are given a higher weight which means that they contribute more to the LSQ solution

Increasing weighting of lower amplitudes: lower amplitude values are given a higher weight which means that they contribute more to the LSQ solution

Do log-linear curve fit:

Converts the exponential expression to a linear expression by logarithmic conversion. The resulting expression can then be fitted by linear regression, which is much faster than exponential curve fits. Care should be taken using this option if input data has one or more datapoints at the noise level since the linearity assumption then will be violated.

Related topics:

T2 / R2 relaxation analysis