Development and Testing of An Automatic Lung IMRT Planning Algorithm

<p>Knowledge-based radiation treatment is an emerging concept in radiotherapy. It</p><p>mainly refers to the technique that can guide or automate treatment planning in</p><p>clinic by learning from prior knowledge. Dierent models are developed to realize</p><p>it, one of which is proposed by Yuan et al. at Duke for lung IMRT planning. This</p><p>model can automatically determine both beam conguration and optimization ob-</p><p>jectives with non-coplanar beams based on patient-specic anatomical information.</p><p>Although plans automatically generated by this model demonstrate equivalent or</p><p>better dosimetric quality compared to clinical approved plans, its validity and gener-</p><p>ality are limited due to the empirical assignment to a coecient called angle spread</p><p>constraint dened in the beam eciency index used for beam ranking. To eliminate</p><p>these limitations, a systematic study on this coecient is needed to acquire evidences</p><p>for its optimal value.</p><p>To achieve this purpose, eleven lung cancer patients with complex tumor shape</p><p>with non-coplanar beams adopted in clinical approved plans were retrospectively</p><p>studied in the frame of the automatic lung IMRT treatment algorithm. The primary</p><p>and boost plans used in three patients were treated as dierent cases due to the</p><p>dierent target size and shape. A total of 14 lung cases, thus, were re-planned using</p><p>the knowledge-based automatic lung IMRT planning algorithm by varying angle</p><p>spread constraint from 0 to 1 with increment of 0.2. A modied beam angle eciency</p><p>index used for navigate the beam selection was adopted. Great eorts were made to assure the quality of plans associated to every angle spread constraint as good</p><p>as possible. Important dosimetric parameters for PTV and OARs, quantitatively</p><p>re</p><p>ecting the plan quality, were extracted from the DVHs and analyzed as a function</p><p>of angle spread constraint for each case. Comparisons of these parameters between</p><p>clinical plans and model-based plans were evaluated by two-sampled Students t-tests,</p><p>and regression analysis on a composite index built on the percentage errors between</p><p>dosimetric parameters in the model-based plans and those in the clinical plans as a</p><p>function of angle spread constraint was performed.</p><p>Results show that model-based plans generally have equivalent or better quality</p><p>than clinical approved plans, qualitatively and quantitatively. All dosimetric param-</p><p>eters except those for lungs in the automatically generated plans are statistically</p><p>better or comparable to those in the clinical plans. On average, more than 15% re-</p><p>duction on conformity index and homogeneity index for PTV and V40, V60 for heart</p><p>while an 8% and 3% increase on V5, V20 for lungs, respectively, are observed. The</p><p>intra-plan comparison among model-based plans demonstrates that plan quality does</p><p>not change much with angle spread constraint larger than 0.4. Further examination</p><p>on the variation curve of the composite index as a function of angle spread constraint</p><p>shows that 0.6 is the optimal value that can result in statistically the best achievable</p><p>plans.</p>

Thesis