soorena62
51 lines
1.5 KiB
Python
51 lines
1.5 KiB
Python
import pysera
|
|
import os
|
|
from workflow.module import Module
|
|
from workflow.io_port import InPort, OutPort, NIFTIImageType, CSVTableType
|
|
# Codes Go Below:
|
|
|
|
|
|
class FeatureExtractor(Module):
|
|
"""
|
|
Pure logical node for radiomics feature extraction.
|
|
No Dagster, no Engine, no state.
|
|
"""
|
|
|
|
def __init__(self):
|
|
super().__init__("FeatureExtractor")
|
|
|
|
self.addInPort(InPort("image", NIFTIImageType()))
|
|
self.addInPort(InPort("mask", NIFTIImageType()))
|
|
|
|
self.addOutPort(OutPort("features", CSVTableType(columns=["name", "value"])))
|
|
|
|
def run(self, context):
|
|
image = context.get_asset_value("FeatureExtractor.image")
|
|
mask = context.get_asset_value("FeatureExtractor.mask")
|
|
|
|
output_dir = "results"
|
|
os.makedirs(output_dir, exist_ok=True)
|
|
|
|
result = pysera.process_batch(
|
|
image_input=image,
|
|
mask_input=mask,
|
|
output_path=output_dir,
|
|
num_workers="auto",
|
|
enable_parallelism=True,
|
|
apply_preprocessing=True,
|
|
categories="all",
|
|
dimensions="1st,2_5d,3d",
|
|
feature_value_mode="REAL_VALUE",
|
|
extraction_mode="handcrafted_feature",
|
|
report="info",
|
|
)
|
|
|
|
df = result.get("features_extracted")
|
|
|
|
features = []
|
|
if df is not None:
|
|
for idx, row in df.iterrows():
|
|
features.append({"name": row[0], "value": row[1]})
|
|
|
|
return {"features": features}
|