2969 - Effectiveness of Photon-Counting CT on Lung Tumor Edge Definition for Target Delineation

Purpose/Objective(s):

Photon-counting CT (PCCT) offers higher spatial resolution and reduced electronic noise than conventional energy-integrating detector CT, but its impact on radiotherapy target delineation remains underexplored. One potential advantage is on identifying the edge of lung tumors. By making use of the improved spatial resolution and a sharper reconstruction kernel, the edge of the tumor can be clearly differentiated from the lung parenchyma. The capability of it was demonstrated to identify some interstitial lung disease. However, a direct comparison for edge detection of tumor is yet to be done.

Materials/Methods:

Four patients with lung tumors underwent simulation on a PCCT, iodine contrast was administrated. Two image series were reconstructed from a single scan within a 500 mm field-of-view: (1) a conventional image set (T3D) using all energy bins, a Qr40 kernel, and 1.0 × 1.0 mm in-plane resolution; and (2) a high-resolution spectral image set using 50 keV Virtual Monoenergetic Images(VMI) with 0.5 × 0.5 mm in-plane resolution and a sharper kernel, Bl64. For each patient, line profiles were drawn across the tumor–lung interface, and the first derivative of the Hounsfield unit (HU) profile was calculated; the maximum absolute derivative at each edge and the HU difference between tumor and adjacent lung parenchyma were recorded for both reconstructions.

Results:

Across all four patients, the spectral high-resolution images showed steeper HU gradients at the tumor boundary and larger tumor–lung HU differences than the conventional T3D reconstructions. In one representative case, the maximum HU gradient for patient 3 was higher on the VMI high-resolution images, indicating a clearer tumor edge relative to the conventional set. Tumor HU values were slightly higher, while image noise remained clinically acceptable, consistent with prior reports that low-keV VMIs can enhance contrast without prohibitive noise penalties. Details were tabulated.

Conclusion:

PCCT’s combination of higher spatial resolution and spectral imaging enables clearer lung tumor edge definition and modestly improved tumor-to-lung contrast for radiotherapy target delineation. Since both the high-resolution edge-enhanced images and the treatment planning CT dataset can be acquired in a single CT simulation, the entire structure set (targets and OARs) can be used for planning without any registration, potentially reducing geometric uncertainty in the radiotherapy workflow

		HU difference between tumor and lung parenchyma (HU)	Maximum value in 1st derivatives of the line profile (HU/mm)
			Proximal end	Distal end
Patient 1	T3D, Qr40	925	411	455
	50 keV, Bl64	958	1317	1731
Patient 2	T3D, Qr40	807	287	299
	50 keV, Bl64	868	922	826
Patient 3	T3D, Qr40	858	439	310
	50 keV, Bl64	910	1901	1740
Patient 4	T3D, Qr40	844	308	314
	50 keV, Bl64	912	1084	978