The Scan class

The pylidc.Scan class holds some (but not all!) of the DICOM attributes associated with the CT scans in the LIDC dataset. These attributes can be used to query the data:

import pylidc as pl

# Query for all CT scans with desired traits.
scans = pl.query(pl.Scan).filter(pl.Scan.slice_thickness <= 1,
                                 pl.Scan.pixel_spacing <= 0.6)
print(scans.count())
# => 31

pid = 'LIDC-IDRI-0078'
scan = pl.query(pl.Scan).filter(pl.Scan.patient_id == pid).first()

Note

Not all of the class attributes are queryable! Some attributes are actually computed properties. You can determine this by, for example, examining the object type for the class:

print(type(pl.Scan.slice_thickness))
# => <class 'sqlalchemy.orm.attributes.InstrumentedAttribute'>
# => ^^^ queryable because it's an sqlalchemy attribute.

# Whereas ...
print(type(pl.Scan.slice_spacing))
# => <type 'property'>
# ^^^ not queryable because it's a computed property

The slice_thickness attribute corresponds to the DICOM attribute (0018,0050), whereas the slice_spacing attribute is computed from the median of spacing between the slices (see pylidc.Scan.slice_zvals()).

A pylidc.Scan object has zero or more pylidc.Annotation objects, which are radiologist annotations of lung nodules found in the scan:

print(len(scan.annotations))
# => 13

Annotation clustering

The scan has 13 annotations, but which refer to the same nodule? This can be determined using the pylidc.Scan.cluster_annotations() method, which uses a distance function to create an adjancency graph to determine which annotations refer to the same nodule in a scan:

nods = scan.cluster_annotations()

print("%s has %d nodules." % (scan, len(nods)))
# => Scan(id=1,patient_id=LIDC-IDRI-0078) has 4 nodules.

for i,nod in enumerate(nods):
    print("Nodule %d has %d annotations." % (i+1, len(nods[i])))
# => Nodule 1 has 4 annotations.
# => Nodule 2 has 4 annotations.
# => Nodule 3 has 1 annotations.
# => Nodule 4 has 4 annotations.

Converting scan image values to NumPy array

The scan can be converted to a NumPy array for image-processing:

vol = scan.to_volume()
print(vol.shape)
# => (512, 512, 87)

print("%.2f, %.2f" % (vol.mean(), vol.std()))
# => -702.15, 812.52

Scan visualization

An interactive GUI can be engaged, with clustered annotations optionally indicated with arrows:

scan.visualize(annotation_groups=nods)

which appears like