stowler

AFNI 3dmaskdump INPUT
------------------------------------------------
Appears to use system rounding rint() to convert coordinate floats to integers: 

If user inputs mm coordinates as 3dmaskdump argument, 3dmaskdump converts them to
integer "index coordinates" using serial calls to three functions:
	1) THD_dicomm_to_3dmm() (from thd_coords.c)
	2) THD_3dmm_to_3dfind() (from thd_coords.c)
	3) rint() (from system's buit-in c math, or from faked version in mri_free.c)

stowler

AFNI 3dmaskdump OUTPUT:
------------------------------------------------
Appears to use implicit casting and a custom log-based algorithm to convert
coordinate floats to integers:

If user requests integer voxel coordinates (not mm), 3dmaskdump creates them
from internal float voxel coordinates (not mm) using functions in multivector.c:
MV_format_fval(), which calls MV_fval_to_char()

e.g., here 3dmaskdump calls for outputting integer voxel coordinate ijk: 

stowler

AFNI controller window user INPUT
------------------------------------------------
Appears to use implicit casting + boundary handling to create integer voxel
coordinates from the user-specified mm coordinates in the "Jump to (xyz)" box:


In afni.c , typical use of AFNI_jumpto_dicom() calls AFNI_jumpto_dicom_OLD(),
which converts user-provided mm coordinates to integer voxel coordinates in
three steps:

stowler

AFNI controller window OUTPUT 
------------------------------------------------
Appears to pull existing integer voxel coordinates from internal integer-based
data structures. Doesn't need conversion from float to integer.

When the AFNI controller window is set to display voxel coordinates for the
current crosshair location (GUI "Set Coord Order"->"Voxel Indexes"), it displays
integer coordinates pulled from internal integer data structures.

from afni.c :

stowler

FSL fslview cursor widget
------------------------------------------------
Appears to use explicit casting to convert float mmm coordinates to integer
voxel coordinates, truncating in the process:

line 266 from fslview's cursorwidget.cpp:

	232 void CursorWidget::mmBoxChanged()
	233 {
	...

stowler

FSL fslmeants INPUT
------------------------------------------------
Appears to round coordinate floats to integers via MISCMATHS::round 
(adding or subtracting .5 then truncating decimal via cast):

If user inputs mm coordinates as fslmeants argument (-c --usemm), fslmeants
rounds to integer voxel coordinates using call to MISCMATHS::round().

From fslmeants.cc:

stowler

FSL fslmeants OUTPUT
------------------------------------------------
If user requests voxel coordinate output (not mm) via fslmeants --showall,
fslmeants doesn't appear to convert floats to integers. It just matrix
transforms internal voxel coordinates (which already lack decimal values due to
their row/slice/column nature) to nifti voxel coordinates.

From fslmeants.cc:

	311               if (showall.value()) {

stowler

$ 3dinfo ~/Downloads/ftp.medimg.emory.edu/NMCardiac/JonNye/PiB_binding_maps/NIFTI/VN040_20110412_LI_rsl.nii
++ 3dinfo: AFNI version=AFNI_2011_12_21_1014 (Dec  1 2012) [64-bit]

Dataset File:    /home/stowler-local/Downloads/ftp.medimg.emory.edu/NMCardiac/JonNye/PiB_binding_maps/NIFTI/VN040_20110412_LI_rsl.nii
Identifier Code: NII_MVGvk3_9itSHbYxclrBKWw  Creation Date: Tue Apr 23 12:23:56 2013
Template Space:  ORIG
Dataset Type:    Anat Bucket (-abuc)
Byte Order:      LSB_FIRST {assumed} [this CPU native = LSB_FIRST]
Storage Mode:    NIFTI
Storage Space:   25,090,560 (25 million) bytes

stowler

#!/bin/bash
#
# CALL AS:        lut_convertToFslview.sh 
#                 (no arguments: input and output files defined below)
#
# CREATED:        20130430 by [email protected]
# LAST UPDATED:   same
#
#
#

stowler

#!/bin/bash

# A short proof-of-concept script for meeting with Pearce and Evan (W, 20130501)
#
# Demonstrates adaptation of existing tools for automated PET->T1->MNI152
# registration. 
#
# INPUT: dicoms for a deidentified participant's T1 and PET sequences.
#
# OUTPUT: PET and T1 nifti volumes nonlinearly registered to MNI152 template (RPI).