Depth of Interaction Sensing for High Resolution Tomography

This direction is merged into another direction: parallax error. Its last version stays readable and citable here.

Version 1 of 110 papers18 conceptsfrozen 2026-10-04

In medical imaging and particle detection, there is a persistent, frustrating trade-off: you can have high sensitivity (catching more particles) or high spatial resolution (seeing fine details), but it is incredibly difficult to have both. This conflict is rooted in the geometry of the detectors themselves.

As researchers like Yuli Wang and Shiva Abbaszadeh have noted, increasing sensitivity usually requires making detector crystals thicker to stop more photons, or moving the detectors closer to the patient. However, both of these actions exacerbate a phenomenon known as parallax error, which blurs the resulting image.

The emerging solution across Positron Emission Tomography (PET), SPECT, and neutron imaging is the implementation of Depth of Interaction (DOI) sensing. By determining exactly where a particle strikes along the thickness of a detector, researchers are breaking the traditional limits of imaging performance.

The Geometry of the Parallax Problem

To understand why DOI is necessary, imagine a ring of detector crystals. If a photon originates in the exact center of the ring and travels outward, it strikes the crystal face perpendicularly. Even if the crystal is very long, the system knows exactly which "pixel" was hit.

However, most photons originate away from the center. These photons strike the crystals at oblique angles. Because the crystals have a finite thickness (often 20mm or more to ensure the photon is actually stopped), a photon might enter at one point but travel deep into the crystal before interacting. Without DOI information, the system assumes the interaction happened at the front face. This uncertainty creates a "blur" in the reconstruction, as the true Line of Response (LOR) could be anywhere within the crystal's volume.

As K. S. Grogg and colleagues explain in their work on the NeuroPET/CT, this parallax effect is most severe at the edges of the field of view. In compact, small-bore scanners designed for the brain or extremities, the angles become so steep that the image quality at the periphery degrades significantly. This same geometric challenge appears in gaseous detectors; Joseph D. Parker demonstrated that the finite depth of gas volumes causes similar resolution degradation in neutron imaging.

How Depth is Measured

If we can measure the depth at which the interaction occurs, we can mathematically correct the parallax. Researchers have developed several clever hardware mechanisms to "see" into the crystal:

  1. Dual-Ended Readout: By placing light sensors (like SiPMs) at both the front and back of a scintillation crystal, the system can compare the relative signal strengths. As Yidong Fu describes in directional neutron detection, the ratio of light reaching each end tells you exactly how far along the bar the interaction happened.
  2. Multi-Layer Crystals: Some systems, like the NeuroPET/CT evaluated by Quanzheng Li, use two or more distinct layers of crystals. By identifying which layer triggered, the system gains a coarse but effective estimate of depth.
  3. Light-Sharing Windows: Siyuan Han and colleagues have developed BGO-based detectors that use specialized windows to control how light spreads between crystals based on depth, allowing for sub-millimeter precision even at the edges of the imaging field.

Enabling New Imaging Architectures

DOI sensing isn't just an incremental improvement; it enables entirely new scanner designs that were previously impossible.

In flat-panel PET systems, Gašper Razdevšek and Andrej Studen show that while long crystals are needed for sensitivity, they normally ruin resolution in the direction perpendicular to the panels. DOI allows these panels to maintain high resolution, which, when combined with ultra-fast timing (50 ps), can compensate for the "missing angles" of a non-ring scanner.

Perhaps the most radical application is the "self-collimating" SPECT system proposed by Qingyang Wei and Yaqiang Liu. Traditional SPECT uses heavy lead collimators to block "stray" photons, which also blocks 99.9% of the useful signal. Their MATRICES architecture uses multiple layers of DOI-capable detectors to track the path of photons without needing to block them, potentially increasing sensitivity by 40 times.

By solving the parallax problem, DOI sensing allows imaging systems to be smaller, more sensitive, and more precise simultaneously, moving high-end molecular imaging from dedicated basement suites to the patient's bedside.

Papers in this version

History & citationVersion 1 · 2026-10-04

Merged into another direction · Permanent link

View frozen version 1

Cite this direction

Latest version (a living page):

@misc{lacuna_depth_of_interaction_sensing_for_high_re_58f4b427,
  title = {Depth of Interaction Sensing for High Resolution Tomography},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/depth-of-interaction-sensing-for-high-resolution-tomography/rdr_58f4b42719454211bda1386d51592a1e},
  note = {Latest version of a living research direction; cite a version for a fixed text. Accessed 2026-11-21},
}

This exact version (version 1, frozen text):

@misc{lacuna_depth_of_interaction_sensing_for_high_re_58f4b427_v1,
  title = {Depth of Interaction Sensing for High Resolution Tomography},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/depth-of-interaction-sensing-for-high-resolution-tomography/rdr_58f4b42719454211bda1386d51592a1e/v/1},
  note = {Version 1, frozen 2026-10-04. Accessed 2026-11-21},
}