Fast and accurate biophotonic simulations for photodynamic cancer therapy treatment planning

光动力癌症疗法（PDT）通过使用一种化合物来治疗癌症，当这种化合物被光激活时，会破坏细胞。通过限制组织暴露在光线下的位置，人们可以用一种比手术侵入性更小的方法摧毁癌细胞，但副作用比传统的化疗或电离放射治疗更少。与这些疗法相比，这缩短了患者的恢复时间，并大大降低了成本。虽然PDT现在被广泛用于治疗皮肤癌，因为它很容易控制哪些组织暴露在光线下，但它很少用于治疗体内的间质癌。这些癌症可以通过PDT治疗，通过静脉注射光敏剂化合物，并通过光纤“光针”照射光激活肿瘤附近的光敏剂。然而，确定这些光探针应该放置在哪里是困难的，并且还不可能在治疗前确定最终的光分布和组织破坏将是什么；这极大地限制了PDT对这些更复杂的癌症的有用性。在这个项目中，我们正在开发一个PDT“治疗计划”系统，该系统既可以预测给定探针放置的光分布，也可以自动建议一个优化的探针放置，从而产生最佳的治疗效果。磁共振成像数据既可以描述肿瘤的位置，也可以描述病人体内不同组织之间的边界，在给定一组光探针的情况下，该系统可以快速而非常准确地模拟光线的流向。这种计算通常非常耗时，但我们将使用称为现场可编程门阵列的特殊计算机芯片来加速计算，使我们的计算效率提高约60倍。有了如此快速的计算，我们可以自动确定许多不同可能探针位置的光分布，并将一组建议的优化探针位置返回给临床医生，以及对肿瘤和周围组织的预期结果。综上所述，该系统可以使PDT治疗对复杂的癌症有效，使一种有前景的新的癌症治疗方法适用于更多的患者，并改善他们的生活。

Photodynamic Cancer Therapy (PDT) treats cancer by using a compound that, when activated by light, destroys cells. By limiting where tissue is exposed to light, one can destroy cancerous cells with a procedure that is less invasive than surgery, but with fewer side-effects than traditional chemotherapy or ionizing radiation therapy. This improves patient recovery times and greatly reduces cost compared to these therapies. While PDT is now widely used to treat skin cancers, where it is easy to control which tissue is exposed to light, it has been less used for interstitial cancers within the body. These cancers can be treated with PDT by administering the photosensitizer compound intraveneously, and activating the photosensitizer only near the tumour by light administered with fiber optic "light needles." However, determining where these light probes should be placed is difficult and it is not yet possible to determine in advance of treatment what the resulting light distribution and hence tissue destruction will be; this greatly limits PDT's usefulness for these more complex cancers.In this project we are developing a PDT "treatment planning" system which will both predict the light distribution that will result from a given probe placement, and automatically suggest an optimized probe placement that will yield the best treatment result. Given MRI data that describes both where the tumour is and the boundaries between the various tissues in a patient's body, this system will rapidly but very accurately simulate where light would go, given a set of light probes. This computation is normally very time-consuming, but we will use special computer chips called Field Programmable Gate Arrrays to accelerate the computation, making our computations approximately 60x more efficient. With such a fast computation, we can automatically determine the light distribution for many different possible probe locations, and return a suggested set of optimized probe locations to a clinician, along with the expected result on the tumour and surrounding tissue. Taken together, this system can make PDT therapy effective for complex cancers, making a promising new cancer therapy applicable to many more patients and improving their lives.