rmsandu / xml-processing Goto Github PK

If multiple registrations have been created at different time points (let's assume by different people) how do we assign them?

Registration belongs to patient hierarchy. One patient can have one or more registrations. A registration is not dependent on a needle trajectory. Issues with ruining the tabular format.

• Calculate area [mm^2] between needle electrodes - implements Shoelace formula
• add it as a new parameter to Excel
• validate the formula for area on paper

when:

• calculating the area of a polygon
• calculating the sum of absolute angles

• clean up the code, move to separate file
• start needle count at 1 for version 2.5
• write to Excel file all stats per lesion, intervention, group

• add patient id from "patient_information.xml" (in older versions it doesn't appear)

• check version 2.5 (where no reference needle is defined)

• check lesion identification algorithm

• Check validation needle

• add patient id

• add patient name

• according to the clinician: only IRE were done. if MWA appears is actually an IRE electrode, since the option for IRE wasn't available in the early versions of CAS-One IR

• add graphical user interface that accepts changeable input params
• copying
• moving
• unzipping
• add input patient name or patient id for selecting the folders.
• rename patient folder with umlauts and other weird characters because they can't be loaded into CAS One Software

• compare with standard ideal of sum for different electrode configurations (3 needles: triangle, 4 needles: square, 5 needles?, 6 needles?)

Extract MWA Needle Information

• TPES
• EntryPoint, TargetPoint
• Type of Needle from the Plan
• which needle for which segmentation
• extract path to segmentation from XML logs (Plan.xml, Ablation_Validation.xml or Validation.xml)
• add the ablator Information
• connect with MWA Needle Database
• get unique patient id from the XML log
• match patient id with patient id from segmentations
• extract date of segmentation from folder name (new CAS version
• add all info to one main excel file (connect everything somehow)
• write to Excel
• fix for ablation_validation.xml new files
• add path to CT plan and CT validation

Or the number of segmentations differs from tumors to ablation.
Scenario 1):
Sometimes the tumor is not visible on the CT scan (only visible on MRI). So it's not possible to segment a tumor if it;s not visible. However, the respective ablation area is visible.
Scenario 2):
No ablation segmentation, but tumor segmentation available. In case of IRE.

• too many variables unwrapped into one structure. problem either with parsing "needles" either with "segmentations"

Failure to open XML logs with German/Swedish umlauts.
These are converted to "weird" characters that cannot be interpreted by Untangle library.
Possible solution:

• open as text, edit the culprit characters.
• parse with untangle

• extract registration matrix from XML
• extract to excel
• flag for registration
• add UID for registration matrices (reg matrix)
• add matrix info to the excel table
• unpack the dict of the pair points

• calculate angles for each electrode pair
• ignore MWA needles
• start needle indexing at 1 (when no reference needle available)
• order lesion nr
• add euclidean distance

• rootfolder
• output file name excel
• flag for Angle between Needles calculation

Unzipping fails because the filepath and filename are too long after appending the CT series UID to XML Validation/Plan file (new version starting 2018).

Because of this it cannot be unzipped, nor opened in Python.

• Update Segmentation Path
• why do we have redundant information? --> union TPEs and Segmentation Paths based on TP
• Update Needle and Lesion Nr for MWA
• Correlate with Lesion Nr (need to extract from RedCAP)
• clean code for writing to Excel
• Validate for IREs

• segmentation path not assigned to existing needle trajectory , create validation check-ups for segmentaitons paths
• assign segmentation to correct trajectory
• segmentations paths overwrite the needle plan&validation trajectory

• parse all TPEs.xlsx together in one Excel file
• calculate metrics on them

• no segmentation present
• no TPEs present
• print which needles need validation
• match lesion number with RedCap electronic records
• needle trajectories planned in different Plan.xml

• do check the spacing of the images & their segmentations
• re-compute the distance metrics (plots)
• still open: where to store the segmentations? Write directly on cochlea
• apply registration matrix on source images
• apply registration matrix on segmentations

• where and Tag available for type = "EG_IRE_Trajectory" or available for type="EG_ATOMIC_TRAJECTORY" (MWA)
• for CAS Version where Trajectory id="0" and so forth can't tell how many lesions based on the XML solely
• reset lesion count when date of intervention changes
• reset lesion count when needle type changes from IRE to MWA
• improve findLesion fct

Generally the IR data of the patients is contained in .ZIP folders
At the moment folders are unzipped manually.
For the future define separate function that creates (Temporary) unzipped folder