Deleted branch from rMITK MITK: bugfix/T29592-Remove3DRegionGrowingFromUserGUide.
- Queries
- All Stories
- Search
- Advanced Search
- Transactions
- Transaction Logs
Advanced Search
May 9 2023
Pushed new branch to rMITK MITK: bugfix/T29592-Remove3DRegionGrowingFromUserGUide.
I have the feeling it is an race condition/timing problem.
I was able to recreate the problem also in release after rapidly clicking arround (at least 50 times, machine gun style). In debug I clicked for minutes without anything appening.
Deleted branch from rMITK MITK: bugfix/T29593-FixViewNavigatorFont.
Pushed new branch to rMITK MITK: bugfix/T29593-FixViewNavigatorFont.
May 8 2023
In T29567#247266, @a178n wrote:Is @wasserth active here? Or we goto GitHub and make an issue over there?
I would also open an issue at GitHub. As you already analyzed the current versions, we could also think of providing a pull request (and have the fork with that changes as back up plan for the release).
May 7 2023
May 6 2023
In T29571#247278, @floca wrote:I can do it, if you want.
It's not an IO error but again the interpolation.
I can do it, if you want.
May 5 2023
Is @wasserth active here? Or we goto GitHub and make an issue over there?
I think it would be more efficient /sufficient to advise @wasserth to use explicitly defined versions for total segmentator dependencies. That is something all projects should regard for productive versions to have a better control on their supply chain.
It seems urllib3 package, a dependency of Totalsegmentator had a version bump yesterday rendering it unusable on MacOS. Its error message clearly shows it as a known issue.
The only solution is to downgrade the version.
This is not an MITK problem, until now we had let TotalSegmentator (implicitly, PyPI) handle the dependency resolution. From this experience, I think we cannot let it happen and take things into our own hands and install the main dependencies ourselves, Currently, we only explicitly handle Pytorch and scipy installation. This should be extended to more packages like urllib3, an HTTP client for Python.
Otherwise, we are always at the risk of TotalSegmentator breaking down in production by no mistake of our own.
Log shows:
TotalSegmentator -i /tmpmitk-AQVb2K/totalseg-in-QOuAM7/EkysIV_000_0000.nii.gz -o /tmpmitk-AQVb2K/totalseg-out-2CpWaL/EkysIV_000.nii.gz --ml --fast /Users/ashis/Library/Application Support/DKFZ/.totalsegmentator/bin
May 4 2023
Deleted branch from rMITK MITK: bugfix/T29563-UpdateSegmentationViewUserGuide.
Pushed new branch to rMITK MITK: bugfix/T29563-UpdateSegmentationViewUserGuide.
Deleted branch from rMITK MITK: bugfix/T29561-FixCrashesIn3DInterpolation.
Pushed new branch to rMITK MITK: bugfix/T29561-FixCrashesIn3DInterpolation.
May 3 2023
Deleted branch from rMITK MITK: bugfix/T29559-itksysProcessVideoRecoder.
Pushed new branch to rMITK MITK: bugfix/T29559-itksysProcessVideoRecoder.
Pushed new branch to rMITK MITK: feature/T29558-totalseg-update.
Deleted branch from rMITK MITK: bugfix/T29556-AddDeleteLabelInstanceToContextMenu.
You should do a compbination (to keep the 4D support).
Hi @floca
The code in UpdatePrepare function for TotalSegmentator has a problem. Not a bug. The labelSet that you modify for the preview preemptively contains all possible labels, not just the ones that are predicted.
void mitk::TotalSegmentatorTool::UpdatePrepare() { Superclass::UpdatePrepare(); auto preview = this->GetPreviewSegmentation(); auto labelset = preview->GetLabelSet(preview->GetActiveLayer()); labelset->RemoveAllLabels(); if (m_LabelMapTotal.empty()) { this->ParseLabelNames(this->GetLabelMapPath()); } auto labelMap = m_LabelMapTotal; const bool isSubTask = (this->GetSubTask() != DEFAULT_TOTAL_TASK); if (isSubTask) { std::vector<std::string> files = SUBTASKS_MAP.at(this->GetSubTask()); labelMap.clear(); mitk::Label::PixelType labelId = 1; for (auto const& file : files) { std::string labelName = file.substr(0, file.find('.')); labelMap[labelId] = labelName; labelId++; } } auto lookupTable = mitk::LookupTable::New(); lookupTable->SetType(mitk::LookupTable::LookupTableType::MULTILABEL); for (auto const& [key, val] : labelMap) { Label::Pointer label = Label::New(key, val); std::array<double, 3> lookupTableColor; lookupTable->GetColor(key, lookupTableColor.data()); Color color; color.SetRed(lookupTableColor[0]); color.SetGreen(lookupTableColor[1]); color.SetBlue(lookupTableColor[2]); label->SetColor(color); labelset->AddLabel(label, false); } }
So, once the segmentation operation is completed and labels are shown in the List widget, all the possible labels are listed out for Confirmation instead of just what was really predicted. This is undesirable.
I think we should only show what label was predicted. For that, I believe, the labelset of the preview segmentation should be set only after its predicted, inside the DoUpdatePreview.
Basically, the lookup table and color setting done should be done lazily & not in UpdatePrepare method.
Pushed new branch to rMITK MITK: bugfix/T29556-AddDeleteLabelInstanceToContextMenu.