Difference between revisions of "Neurosurgery wishlist"
(New page: 1. Better editor - "magic wand" like Photoshop to select nearby and *connected* pixels with similar intensity - sensitivity controls for the levelset threshold thing - 2. Better segmenta...) |
|||
| Line 1: | Line 1: | ||
1. Better editor | 1. Better editor | ||
| + | |||
- "magic wand" like Photoshop to select nearby and *connected* pixels with similar intensity | - "magic wand" like Photoshop to select nearby and *connected* pixels with similar intensity | ||
| + | |||
- sensitivity controls for the levelset threshold thing | - sensitivity controls for the levelset threshold thing | ||
| + | |||
- | - | ||
2. Better segmentation | 2. Better segmentation | ||
| + | |||
- EMSegmenter is nice, but it is time-consuming and too involved for basic things. Very often all I want is a basic but presentable rendering of a subject-specific cortex. As yet, there is no easy and quick way to do this. BrainLab and GE both have quick surface generation algorithms which work reasonably well based on one or a few intensity seed points. | - EMSegmenter is nice, but it is time-consuming and too involved for basic things. Very often all I want is a basic but presentable rendering of a subject-specific cortex. As yet, there is no easy and quick way to do this. BrainLab and GE both have quick surface generation algorithms which work reasonably well based on one or a few intensity seed points. | ||
3. Tool/UI requests | 3. Tool/UI requests | ||
| + | |||
- Getting the crosshairs squared away and working right is CRUCIAL. I cannot count the number of times a surgeon has asked me for this function. | - Getting the crosshairs squared away and working right is CRUCIAL. I cannot count the number of times a surgeon has asked me for this function. | ||
| + | |||
- need ability to measure angles and distances | - need ability to measure angles and distances | ||
4. Usability things | 4. Usability things | ||
| + | |||
- I'm not sure exactly how, but would like to find a better way to move things around in 3D. The best I can think of so far is rotational motion along some sort of level set, either of a structural image, or a distance function from some pre-defined "central feature". Pushing a (arrow?) key would move in and out.. Maybe this would be less intuitive than the current method of "rotate plane" move, "rotate plane" move, etc... But maybe it would work? | - I'm not sure exactly how, but would like to find a better way to move things around in 3D. The best I can think of so far is rotational motion along some sort of level set, either of a structural image, or a distance function from some pre-defined "central feature". Pushing a (arrow?) key would move in and out.. Maybe this would be less intuitive than the current method of "rotate plane" move, "rotate plane" move, etc... But maybe it would work? | ||
Revision as of 00:24, 27 November 2008
Home < Neurosurgery wishlist1. Better editor
- "magic wand" like Photoshop to select nearby and *connected* pixels with similar intensity
- sensitivity controls for the levelset threshold thing
-
2. Better segmentation
- EMSegmenter is nice, but it is time-consuming and too involved for basic things. Very often all I want is a basic but presentable rendering of a subject-specific cortex. As yet, there is no easy and quick way to do this. BrainLab and GE both have quick surface generation algorithms which work reasonably well based on one or a few intensity seed points.
3. Tool/UI requests
- Getting the crosshairs squared away and working right is CRUCIAL. I cannot count the number of times a surgeon has asked me for this function.
- need ability to measure angles and distances
4. Usability things
- I'm not sure exactly how, but would like to find a better way to move things around in 3D. The best I can think of so far is rotational motion along some sort of level set, either of a structural image, or a distance function from some pre-defined "central feature". Pushing a (arrow?) key would move in and out.. Maybe this would be less intuitive than the current method of "rotate plane" move, "rotate plane" move, etc... But maybe it would work?
