Agentic AI Reasoning and Learned Sampling for Task-Aware Projection Selection in Sparse-View CT: Predicting Hemodynamics

Mahmoud Elhadidy; Roshan M. D'Souza; Amirhossein Arzani

Figure 1: Overview of the study. (a) Data generation: an idealized aneurysm geometry gives a pulsatile velocity field, which drives an advection–diffusion simulation of contrast concentration. The concentration field is voxelized and fan-beam sinograms are synthetically generated. The training dataset contained distinct geometry and flow rates. (b) Prediction from a fixed sparse view set: the predictor

f_{\theta}:\mathbb{R}^{N_{d}\times N_{v}}\rightarrow\mathbb{R}^{M}

maps the sparse-view sinogram directly to the projected time-averaged wall shear stress (TAWSS) on the vessel wall, where

M

is the number of predicted outputs as introduced in Sec. 2.3 . (c) Prediction with learned view selection: the acquired subset

\mathcal{S}

of the

N_{v}

candidate views is optimized together with the predictor model

f_{\theta}

, guided by the task loss.

Figure 2: Five mechanisms for choosing which

K

views of the

N_{v}

candidates are selected. Each row gives the defining equations on the left, the resulting selection in the center, and a similar predictor

f_{\theta}

, implemented as an FNO, on the right. (a) Uniform sampling:

K

equispaced angles, fixed before training. (b) Golden-angle sampling: fixed

137.5^{\circ}

increments. (c) Gumbel–Softmax selection (Gumbel). (d) Soft top-

K

selection. (e) Agentic AI selection (Agentic).

Figure 3: Prediction of spatially resolved TAWSS at different view budgets

K

. (a) Test Rel-

L_{2}

error across five runs. Boxes span the mean

\pm 1

standard deviation, center lines indicate the mean, and whiskers span the mean

\pm 2

standard deviations. Stars mark the lowest mean error at each budget. (b) Gap to the best method at each budget, expressed as Cohen’s

d

: the difference in mean error divided by the pooled standard deviation of the two methods,

d=(\bar{\varepsilon}_{i}-\bar{\varepsilon}_{\mathrm{best}})/\sqrt{(s_{i}^{2}+s_{\mathrm{best}}^{2})/2}

. The best method at each budget is marked best . (c) Mean error normalized by each method’s mean error at

K=90

, showing the relative increase as the view budget decreases. (d) Methods ranked by mean error at each budget, with rank 1 indicating the lowest error. At

K=90

, Agentic and Soft top-

K

are tied. Agentic is shown as rank 1 solely for visualization purposes.

Method	$K=90$	$K=45$	$K=15$	$K=2$	$K=1$
Uniform	$0.0391\pm 0.0056$	$0.0401\pm 0.0052$	$0.0432\pm 0.0069$	$0.1547\pm 0.0217$	$0.1925\pm 0.0165$
Golden	$0.0386\pm 0.0054$	$0.0413\pm 0.0061$	$0.0439\pm 0.0064$	$0.0706\pm 0.0090$	$0.1924\pm 0.0167$
Gumbel	$0.0345\pm 0.0033$	$0.0391\pm 0.0073$	$0.0416\pm 0.0058$	$0.0738\pm 0.0289$	$0.1660\pm 0.0435$
Soft top- $K$	$\mathbf{0.0343\pm 0.0045}$	$0.0348\pm 0.0036$	$0.0407\pm 0.0062$	$0.0554\pm 0.0122$	$0.0623\pm 0.0129$
Agentic	$\mathbf{0.0343\pm 0.0052}$	$\mathbf{0.0347\pm 0.0063}$	$\mathbf{0.0384\pm 0.0051}$	$\mathbf{0.0544\pm 0.0065}$	$\mathbf{0.0588\pm 0.0043}$

Table 1: Field prediction errors (mean

\pm

standard deviation across five runs). Bold indicates the lowest mean error at each view budget. At

K=90

, Agentic and Soft top-

K

have equal mean error to four decimal places and are both marked in bold. At

K=1

, Uniform and Golden use the same view, and their small differences arise from non-deterministic GPU arithmetic during training.

Figure 4: Predicted spatially resolved TAWSS for two test cases, shown for all five methods at view budgets

K

. Each panel gives the prediction with the Rel-

L_{2}

error of that case below it, alongside the ground truth. The cases were selected at random from the test set and the per-case errors therefore differ from the run-averaged values in Table 1 .

Figure 5: Spatially averaged TAWSS errors for view budgets

K

. Panels follow Fig. 3 , with means and standard deviations computed over five runs. In panel (a), the error axis is logarithmic, so lower whiskers that fall below the lowest value shown, including those where the mean minus two standard deviations is negative, are truncated at the bottom of the axis.

Method	$K=90$	$K=45$	$K=15$	$K=2$	$K=1$
Uniform	$0.0264\pm 0.0140$	$\mathbf{0.0177\pm 0.0061}$	$\mathbf{0.0209\pm 0.0092}$	$0.0985\pm 0.0526$	$0.1540\pm 0.0346$
Golden	$\mathbf{0.0166\pm 0.0061}$	$0.0215\pm 0.0053$	$0.0218\pm 0.0081$	$0.0465\pm 0.0186$	$0.1545\pm 0.0317$
Gumbel	$0.0210\pm 0.0067$	$0.0242\pm 0.0056$	$0.0294\pm 0.0116$	$0.0322\pm 0.0087$	$0.0545\pm 0.0106$
Soft top- $K$	$0.0260\pm 0.0127$	$0.0221\pm 0.0056$	$0.0285\pm 0.0114$	$0.0404\pm 0.0134$	$\mathbf{0.0435\pm 0.0176}$
Agentic	$0.0173\pm 0.0063$	$0.0210\pm 0.0054$	$0.0249\pm 0.0049$	$\mathbf{0.0320\pm 0.0075}$	$0.0446\pm 0.0112$

Table 2: Scalar (spatially averaged) prediction errors (mean

\pm

std over 5 runs). Bold marks the best value in each column. At

K=1

, Uniform and Golden use the same view, and their small differences arise from non-deterministic GPU arithmetic during training.

Figure 6: Views selected by each method for the spatially resolved target at

K=45

, drawn over the contrast concentration volume at

t=2

s. Each ray marks one acquired projection angle and the outline shows the full

360^{\circ}

candidate ring of

N_{v}=180

views spaced

2^{\circ}

apart. The first five columns are the five runs, each labeled with its test Rel-

L_{2}

. The last column pools the runs and is labeled with the mean

\pm

standard deviation over them. In this last column, the rays are colored by how many of the five runs selected each view (color scale at the right).

Figure 7: Views selected by each method for the spatially resolved target at

K=15

, drawn over the contrast concentration volume at

t=2

s. Layout as in Fig. 6 .

Figure 8: Views selected by each method for the spatially resolved target at

K=2

, drawn over the contrast concentration volume at

t=2

s. Layout as in Fig. 6 .

Figure 9: Angular structure of the learned masks for Gumbel, Soft top-

K

and Agentic with five runs per target and budget (60 masks per target at

K\leq 45

). (a) Percentage of masks that place at least one view within

\pm 5^{\circ}

of each angle, for

K\leq 45

. The angle axis spans the full

360^{\circ}

gantry rotation. Shaded bands mark

80^{\circ}

–

160^{\circ}

and

240^{\circ}

–

280^{\circ}

. (b) Fraction of a mask that reappears at the next higher budget, averaged over runs and both targets.

Figure 10: Reasoning of the Agentic search over all 390 scored trials from 50 runs. Design concepts are coded by keyword match on the stated candidate construction and a trial may name more than one concept. Each concept is defined in Table 3 . (a) Trials naming each concept. (b) For each concept, the ratio of the trials that were finally chosen to the trials that proposed it. (c) Share of trials by role at each budget, for the spatially resolved (R) and spatially averaged (A) targets. Roles are the agent’s own labels: exploit refines a construction that already worked in that run, explore tests an unproven construction or a deliberate control, and wildcard is an unconventional idea outside the current line of reasoning, which the protocol required each run to attempt. (d) The predecessors each trial declares:

41\%

build only on earlier trials in the same run,

31\%

on both earlier trials and separate analyses,

18\%

on analyses alone, and

9\%

declare none.

Concept	Proposed	Chosen	What it means
Uniform / grid	58	1	Views placed at equal angular spacing, optionally rotated by an offset. The standard sparse-view construction, used here mostly as a starting point.
Block / contiguous	57	10	Views drawn from one or more contiguous angular bands rather than spread over the full range, including bands that wrap past $360^{\circ}$ .
Control / baseline	46	2	A candidate proposed to test a claim rather than to win, such as a deliberately poor placement expected to fail.
Ridge / linear proxy	45	4	Views ranked by a cheap supervised surrogate, typically a ridge or least-squares fit from single views to the target.
Coverage / spread	42	8	Placement driven by an explicit geometric spacing criterion such as farthest-point, stratified or low-discrepancy sampling.
Local refinement	35	7	Small perturbations of an existing mask: shifting a view by a few degrees, jittering, or swapping in an angular neighbor.

Table 3: The design concepts coded in Fig. 10 . A trial is assigned a concept when its stated candidate construction contains one of the listed keywords, so a single trial may carry several concepts. Proposed is the number of trials naming the concept and chosen the number of those that became a run’s final used mask.

Figure 11: How the agents move through design concepts, using the same keyword coding as Fig. 10 over 390 trials from 50 independent searches. Concepts are ordered by their typical position in a run, and both panels share that order. (a) Percentage of trials using each concept at each trial number, with the eighth and later trials pooled. Uniform sampling dominates the first trial, while local refinement and sensitivity checks appear only from the fourth trial onward. (b) For each concept on the left, the three concepts the agent most often chooses in the next trial, ordered from most to least common and colored by that share. A switch is a concept used in the next trial but not in the current one.

Figure 12: Search trajectories of the Agentic view selector: five independent search agents per panel, one column per view budget. (a) Validation error of the candidate proposed at each trial for spatially resolved TAWSS. (b) The lowest error reached up to that trial, i.e., the cumulative minimum of panel (a). (c) , (d) The same two views for spatially averaged TAWSS. Rings mark the trial each agent finally selected.

Method	Selection (s)	Predictor training (s)	Total (min)	Peak GPU memory (MB)
Uniform	0	892.5	14.9	561
Golden	0	892.5	14.9	561
Gumbel	In-train	895.7	14.9	562
Soft top- $K$	In-train	1005.3	16.8	1053
Agentic	570 †	892.5	24.4	617 ∗

Table 4: Computational cost of the view selection methods for the spatially resolved TAWSS at

K=45

.

Figure 13: Robustness to additive sinogram noise for spatially resolved TAWSS. (a)

K=45

views. (b)

K=2

views. Bars are the mean and whiskers

\pm\,\sigma

over five runs. The shaded region marks Rel-

L_{2}

>0.5

, where the TAWSS predictor fails.

Figure 14: Filtered back-projection reconstructions from the mask each method selected for two

y

slices of the same case at

t=2

s. Rows are the five view-selection methods and columns the view budget

K

. All panels are cropped to the object and the concentration is normalized to

0

–

1

within each slice. The number under each panel is the peak signal-to-noise ratio (PSNR) against that slice’s reference.

Agentic AI Reasoning and Learned Sampling for Task-Aware Projection Selection in Sparse-View CT: Predicting Hemodynamics

Abstract

Figures & tables

Computed Tomography

LUCID: Learned Undersampling-Adaptive Consistency-Guided Inference with Deterministic Flow Matching for Sparse-View CT Reconstruction

SONAR: A Structure-Consistent Neural Operator for Null-Space-Aware Sparse View CT Reconstruction

Soft Tuy-Completeness for Robust Projection Selection in Cone-Beam CT

Abstract

Computed Tomography

Explore similar work

LUCID: Learned Undersampling-Adaptive Consistency-Guided Inference with Deterministic Flow Matching for Sparse-View CT Reconstruction

SONAR: A Structure-Consistent Neural Operator for Null-Space-Aware Sparse View CT Reconstruction

Soft Tuy-Completeness for Robust Projection Selection in Cone-Beam CT