Open gdevenyi opened 7 years ago
It's funny you would say this ... in the maget_mask
function we do voxel_vote
the masks and we're thinking of switching to max
, because this sometimes cuts off some voxels of the brain. This might indeed work better after a mincANTS nonlinear registration than in maget_mask
, though.
From: Gabriel A. Devenyi [notifications@github.com] Sent: January 24, 2017 2:42 PM To: Mouse-Imaging-Centre/pydpiper Cc: Subscribed Subject: [Mouse-Imaging-Centre/pydpiper] pydpiper 2.0.5 MBM nlin_mask maybe should be voxel_voted? (#315)
Rather than "-max"ing the candidate masks, perhaps voxel_voting the mask would result in a tighter mask while still being accurate?
— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHubhttps://urldefense.proofpoint.com/v2/url?u=https-3A__github.com_Mouse-2DImaging-2DCentre_pydpiper_issues_315&d=DQMCaQ&c=Dvjge31PR3JZstzk2paJYiflTxRSxp35GBhsbHdd2Zw&r=v3Chm4sCUETLltY6-53m2uQfL8HQ7l79D29BO0w4Fso&m=31eqCoC3nvq9Y6UD0Vzgo3et4T4gtoH1rEUyOOrc9YE&s=lbnSA7fmIlt3hNSs1r3TJcJ-0cpPZRRPFzQhi4AGbys&e=, or mute the threadhttps://urldefense.proofpoint.com/v2/url?u=https-3A__github.com_notifications_unsubscribe-2Dauth_AAKBNAmV6ZMf-5Fa2jQs83ngoQdwUw4Qfzks5rVlQcgaJpZM4Lsspp&d=DQMCaQ&c=Dvjge31PR3JZstzk2paJYiflTxRSxp35GBhsbHdd2Zw&r=v3Chm4sCUETLltY6-53m2uQfL8HQ7l79D29BO0w4Fso&m=31eqCoC3nvq9Y6UD0Vzgo3et4T4gtoH1rEUyOOrc9YE&s=9k8jE6G7bHhIGhDRp5Zk_7uuFvbkPz-uHLuCPLWF7z0&e=.
This e-mail may contain confidential, personal and/or health information(information which may be subject to legal restrictions on use, retention and/or disclosure) for the sole use of the intended recipient. Any review or distribution by anyone other than the person for whom it was originally intended is strictly prohibited. If you have received this e-mail in error, please contact the sender and delete all copies.
So, I have a strong argument against -max.
We just had one subject go off the rails in twolevel. Since twolevel masking is subject specific MAGeT's voting didn't save us. Its masking error propagated to the final mask because of "-max"
Maybe dilate the final mask after voxel vote to include the missing voxels?
I've also had this happen recently. We could add dilation of a configurable number of voxels (or whatever) and possibly also some voting threshold so you could interpolate between majority vote and 'max'.
From: Gabriel A. Devenyi notifications@github.com Sent: November 23, 2017 4:19:29 PM To: Mouse-Imaging-Centre/pydpiper Cc: Benjamin Darwin; Comment Subject: Re: [Mouse-Imaging-Centre/pydpiper] pydpiper 2.0.5 MBM nlin_mask maybe should be voxel_voted? (#315)
Maybe dilate the final mask after voxel vote to include the missing voxels?
— You are receiving this because you commented. Reply to this email directly, view it on GitHubhttps://urldefense.proofpoint.com/v2/url?u=https-3A__github.com_Mouse-2DImaging-2DCentre_pydpiper_issues_315-23issuecomment-2D346698816&d=DwMCaQ&c=Sj806OTFwmuG2UO1EEDr-2uZRzm2EPz39TfVBG2Km-o&r=WbPKw40NU3g_RTKn7pWL3cSAdk6QRKr3kMreWPZzNcg&m=NBgUzI0UklzRqNW0gGzJlkENV3iLVYJodnOExy7t2Us&s=Ul3fcY6P5-zvzcxvbjV1bMrA0jsP-uXg1XJWsoYwT4c&e=, or mute the threadhttps://urldefense.proofpoint.com/v2/url?u=https-3A__github.com_notifications_unsubscribe-2Dauth_AAKBNE1srkK2h1av8dgzsnnsACdgKQpzks5s5eFhgaJpZM4Lsspp&d=DwMCaQ&c=Sj806OTFwmuG2UO1EEDr-2uZRzm2EPz39TfVBG2Km-o&r=WbPKw40NU3g_RTKn7pWL3cSAdk6QRKr3kMreWPZzNcg&m=NBgUzI0UklzRqNW0gGzJlkENV3iLVYJodnOExy7t2Us&s=-iaOvinlULG-wqTYVALYeJ_xsTx2Eu7k_eUoEwkzKx8&e=.
This e-mail may contain confidential, personal and/or health information(information which may be subject to legal restrictions on use, retention and/or disclosure) for the sole use of the intended recipient. Any review or distribution by anyone other than the person for whom it was originally intended is strictly prohibited. If you have received this e-mail in error, please contact the sender and delete all copies.
Just following up on my various issues, I totally agree with the choice to "-max" vote the maget masks. An under segmentation at that stage propagates down through the model build in a bad way.
Rather than "-max"ing the candidate masks, perhaps voxel_voting the mask would result in a tighter mask while still being accurate?