Introduction

IMPROMPTU, which stands for IMage PROcessing Module for the Prototyping, Testing, and Utilization of image analysis processes, is a software package for multi-dimensional image processing.

With IMPROMPTU, a user can construct an arbitrary sequence of operations to run automatically on an input volume. This sequence, referred to as a process, can be made up of operations selected from any of several categories. Each category, which represents one general type of image-processing operation, contains a number of functions.

IMPROMPTU can run in window mode or batch mode. Window mode allows a user to interact with IMPROMPTU via popdown menus. In batch mode, IMPROMPTU is run using UNIX shell-commands. Later sections will give greater detail on these modes. Below is a list of the main attributes of IMPROMPTU:

1. The input is a volume of 8-bit voxel. The input can be one slice (i.e., a 2-D image) if you wish. This input can be read in with IMPROMPTU from disk.
2. The output of processing is written into the memory space occupied by the input volume (provided the user specifies this). It can also be written into a specified 8-bit output volume-database file. It can also interact with VIDA's shared memory.
3. Measurements of various volume parameters can be generated. As examples, gray-level histograms and a volume's global gray-scale standard deviation can be generated. Such outputs are made by creating new files or by printing messages in IMPROMPTU's main menu.
4. Any arbitrary process, within the confines of functions available, can be set up. A process can be saved and used repeatedly. Processing can be done 1-D, 2-D, or 3-D (depending on the function). Also, text files, containing the parameter values of operations in a process can be generated.
5. UNIX script files, which can be run in batch mode and contain shell commands for IMPROMPTU runs on different input volumes and different processes, can be set up.
6. The user can select a prespecified portion of a volume to process.
7. The programmer can add new functions and categories of functions to IMPROMPTU.

Subsequent sections of this report describe the operation of IMPROMPTU, the functions within IMPROMPTU, and the software structure of IMPROMPTU.

• Filter
  • ``Sigma'' -- sigma filter
  • ``maximum homogeneity'' -- the maximum-homogeneity filter
  • ``Symmetric_Nearest_Neighbor'' -- symmetric nearest-neighbor mean filter
  • ``Salt_and_Pepper_Removal'' -- remove salt and pepper (spike) noise
  • ``Anisotropic_Diffusion'' -- anisotropic diffusion filter
  • ``4D_Filtering'' -- 4D mean filter and grayscale inversion
  • ``Median'' -- a median filter
  • ``Average_Maximum'' -- average/maximum filter
  • ``Average_Minimum'' -- average/minimum filter
  • ``Lowpass'' -- lowpass filter
• Morphology
  • ``Binary_Erosion'' -- binary erosion
  • ``Binary_Dilation'' -- binary dilation
  • ``Binary_Opening'' -- morphological binary opening
  • ``Binary_Closing'' -- morphological binary closing
  • ``Minimum'' -- local minimum
  • ``Maximum'' -- local maximum
  • ``Grayscale_Erosion'' -- grayscale erosion
  • ``Grayscale_Dilation'' -- gray-scale dilation
  • ``Grayscale_Opening'' -- morphological grayscale opening
  • ``Grayscale_Closing'' -- morphological grayscale closing
  • ``Conditional_Dilation'' -- conditional dilation
  • ``4D_Morphology'' -- 4D mathematical morphology functions
  • ``tophat transform'' -- detect peaks
  • ``Ultimate_Erosion'' -- ultimate erosion
• Topology
  • ``3D_conn_comp'' -- three-dim. connected components analysis
  • ``2D_conn_comp'' -- two-dim. connected components analysis
  • ``Cavity_Deletion'' -- delete interior cavities
  • ``Homotopic_Thinning'' -- thin an input volume
  • ``Homotopic_Thickening'' -- conditional homotopic thickening
  • ``Segement_Generation'' -- improve skeletonization
• Segmentation
  • ``Hysteresis_Threshold'' -- thresholding using hysteresis
  • ``Seeded_Hysteresis_Threshold ''---Hysteresis Thresholding using a biopsy image
  • ``Region_growing'' -- segment a volume using region growing
  • ``Cue-based_Relaxation_Labeling'' -- segment an image using relaxation labeling
  • ``Cue-based_Relaxation_Labeling'' -- segment an image using relaxation labeling
  • ``Cue-based_Relaxation_Labeling'' -- segment an image using relaxation labeling
  • `Statistical_Relaxation''---Relaxation labeling using manually-set statistics of image regions
  • ``morphological_gradient'' -- morphological gradient
  • ``Sobel edge operator'' -- Sobel edge operator
  • ``Kirsch edge operator'' -- Kirsch edge operator
  • ``border of binary vol'' -- find borders in a binary volume
  • ``nonmax suppression'' -- non-maximum suppression
  • ``3D_Adaptive_thresholding'' -- automatically threshold a 3-D volume
• Manipulation
  • ``copy vol1 to vol2'' -- copy one volume into another
  • ``two-volume ops'' -- pointwise functions combining 2 volumes
  • ``threshold'' -- intensity thresholding
  • ``Gray-scale inversion'' -- Gray-scale inversion
  • ``histogram'' -- compute a histogram of a volume
  • ``3-D histogram'' -- compute a 3-D histogram
  • ``zero hi voxels'' -- zero out bright voxels
  • ``zero lo voxels'' -- zero out dim voxels
  • ``complement'' -- complement a volume
  • ``min.-bounding cuboid'' -- finding object bounds
  • ``volume properties'' -- Compute various properties of a volume
  • ``3D_Expand'' -- generate an expanded (zoomed) version of an image
  • ``3D_Shrink'' -- generate a shrunken (subsampled) version of an image
  • ``save largest region'' -- extract largest region from segmentation volume
  • ``Flip'' -- Flip a 3D Image
  • ``Resize'' -- Save a resized 2D/3D Image
  • ``Range Labelling'' -- Set a range of gray values to one value
  • ``Level Slicing'' -- pull out a range of gray values
  • ``add Gaussian noise'' -- add Gaussian noise to a volume
  • ``umbra'' -- compute the umbra of a volume
  • ``Zero_Padding'' -- pads an image with zeroes.
  • ``Image_Masking'' -- masks out of given region.
• Measurement
  • ``two-vol. comparison'' -- find differences between two volumes
  • ``volume statistics'' -- find properties of a segmented volume
  • ``demask region'' -- identifies segmented regions within a volume
• Turnkey
  • ``LV Extract'' -- extract left ventricular chambers
  • ``solid-blob extract'' -- extract bright solid blobs
  • ``object separation'' -- separate connected objects
  • ``region isolation'' -- extract a large bright region
  • ``fill concave gaps'' -- fill concave gaps in objects
  • ``solid-blob test'' -- run tests on solid-blob extract
• Input/Output
  • ``Display'' -- display a volume in a graphics window
  • ``Load'' -- load a volume into memory
  • ``Save -- write a volume to disk
• In_Progress
  • ``diff. of Gaussians'' -- difference of Gaussians