In order to deeply analyze the application of CT images based on artificial intelligence algorithm in clinical treatment of AIDS patients with gastric cancer, and to provide reference for intervention of AIDS patients with gastric cancer, a total of 100 AIDS patients with gastric cancer were included as the research objects. The patients with CD4+ T lymphocyte count less than 200 cells/µL were in the control group (50 cases), whereas those higher than 200 cells/µL were in the experimental group (50 cases). All the patients underwent 64-slice spiral CT scanning. An improved algebraic reconstruction technology (ART) under L0 algorithmic approach (L0-ART) was proposed, and compared with the total variation (TV), filtered back projection (FBP), and weighted total variation (WTV) models. The standard deviation (STD) and average processing time of the L0-ART algorithm were significantly lower than those of the TV, FBP, and WTV algorithms ( P  < 0.05). The operation time of the experimental group was longer than that of the control group; the intraoperative blood loss, the diameter of the surgical wound, the time of first farting, the length of hospital stays, and the incidence of adverse reactions were all greatly lower than those of the control group ( P  < 0.05). Postoperatively, the total adipose tissue (TAT), visceral adipose tissue (VAT), and subcutaneous adipose tissue (SAT) in the abdominal area were higher in the experimental group than those in the control group ( P  < 0.05). In conclusion, the improved L0-ART algorithm proposed in this study had an excellent processing effect on CT images with a clinical promotion value. Patients with CD4+ T lymphocytes over 200 cells/µL had better surgical outcomes and prognosis than those with less than 200 cells/µL.