Classification of Chaotic Squeak and Rattle Vibrations by CNN Using Recurrence Pattern

The chaotic squeak and rattle (S&R) vibrations in mechanical systems were classified by deep learning. The rattle, single-mode, and multi-mode squeak models were constructed to generate chaotic S&R signals. The repetition of nonlinear signals generated by them was visualized using an unthresholded recurrence plot and learned using a convolutional neural network (CNN). The results showed that even if the signal of the S&R model is chaos, it could be classified. The accuracy of the classification was verified by calculating the Lyapunov exponent of the vibration signal. The numerical experiment confirmed that the CNN classification using nonlinear vibration images as the proposed procedure has more than 90% accuracy. The chaotic status and each model can be classified into six classes.

MET-4 Clinical investigation of the cases recurred as dissemination after postoperative local irradiation for metastatic brain tumors

Last year, the authors examined the outcome of the patients with metastatic brain tumor (MBT)treated by whole-brain irradiation (WBRT)or local irradiation (LRT)after surgery. As a result, it was shown that the overall survival (OS) was same but the recurrence pattern was different. Furthermore, it was shown that there were some cases with disseminated recurrence in the LRT group. One year has passed, cases showing disseminated recurrence after LRT were examined. The subjects were 28 patients for whom LRT was selected as post-surgical irradiation since December 2017, with an average age of 66.2 years and a male-female ratio of 19: 9. Non-small cell lung cancer was the most in 17 cases. During the observation period, recurrence was observed in 12 cases, new outbreaks at other sites in 8 cases, disseminated recurrence in 4 cases, and no local recurrence. There was no clear difference in kinds of carcinoma and removal fashion between disseminated recurrence cases and other cases. Disseminated recurrence occurred between 3–10 months after surgery, 2 presented with headache, 1 with convulsions, 1 confirmed during follow-up of images, and all underwent WBRT. The lesions shrank after irradiation, but they were easy to re-grow, and the prognosis was poor. On the other hand, 10 cases died in 24 cases other than disseminated recurrence, but all cases died of primary cancer. Although LRT after surgery is non-inferior to WBRT in terms of OS and has the advantage of maintaining cognitive function, this study shows that there is a considerable risk of disseminated recurrence regardless of the removal fashion or kinds of carcinoma. It was also shown that prognosis after disseminated recurrence was poor. It is desirable to select postoperative irradiation after explaining the recurrence pattern, and when LRT is adopted, it is necessary to consider setting a short observation interval immediately after irradiation.

Pattern of Recurrence of Glioblastoma Versus Grade 4 IDH-Mutant Astrocytoma Following Chemoradiation

Purpose To quantitatively compare the recurrence pattern of glioblastoma (IDH-wild type) versus grade 4 IDH-mutant astrocytoma (herein referred to as wtIDH and mutIDH, respectively) following primary chemoradiation. Methods Twenty-two wtIDH and 22 mutIDH patients matched by sex, extent of resection, and corpus callosum involvement were enrolled. The recurrent gross tumor volume (rGTV) was compared with both the gross tumor volume (GTV) and clinical target volume (CTV) from radiotherapy planning. Failure patterns were quantified by the incidence and volume of the rGTV outside the GTV and CTV, and positional differences of the rGTV centroid from the GTV and CTV. Results The GTV was smaller in wtIDH compared to the mutIDH group (mean±SD: 46.5±26.0 cm3 v. 72.2±45.4 cm3, p=0.026). The rGTV was 10.7±26.9 cm3 and 46.9±55.0 cm3 smaller than the GTV for the same groups (p=0.018). The rGTV extended outside the GTV in 22 (100%) and 15 (68%) (p=0.009) of wtIDH and mutIDH patients, respectively; however, the volume of rGTV outside the GTV was not significantly different (12.4±16.1 cm3 vs. 8.4±14.2 cm3, p=0.443). The rGTV metrics extending outside the CTV was not different between the groups. The rGTV centroid was within 5.7 mm of the closest GTV edge for 21 (95%) and 22 (100%) of wtIDH and mutIDH patients, respectively. Conclusion The rGTV extended beyond the GTV less often in mutIDH patients, suggesting limited margin radiotherapy could be beneficial in this group. The results support the study of small margin adaptive radiotherapy per the ongoing UNITED MR-Linac 5 mm CTV trial (NCT04726397).

Radiotherapy for Vaginal Recurrences of Cervical Cancer in Patients After Prior Surgery: Analysis of Effect and Prognostic Factors

ObjectiveThe role of salvage radiotherapy (RT) in the treatment for vaginal recurrence of cervical cancer in patients after prior surgery remains controversial. The aim of this study was to evaluate the efficacy and toxicity of salvage RT and explore prognostic factors associated with the survival after recurrence.MethodsPatients with cervical cancer, treated for vaginal recurrences at Peking Union Medical College Hospital between July 2011 and November 2019, were identified. All the patients underwent prior surgery for primary tumor and received salvage RT including external beam radiotherapy (EBRT), brachytherapy (BT), or both. The irradiation field and dose depended on the conditions of patients. Recurrence patterns were classified into four categories according to the site of recurrence. Prognostic factors on the overall survival (OS), progression-free survival (PFS), and local control (LC) were analyzed, and late toxicity was evaluated.ResultsA total of 141 patients were included in the analysis, with a median follow-up time of 40.8 months. The estimated 5-year OS, PFS, and LC rates were 81%, 75%, and 87%, respectively. In multivariate analysis, endovaginal recurrence and no irradiation history were favorable prognostic factors associated with OS (all p < 0.05), PFS (all p < 0.05), and LC (all p < 0.05). The area under the receiver operating characteristic (ROC) curve (AUC) of the recurrence pattern is larger than the stage of primary tumor (0.734 vs. 0.670).ConclusionsRT was an effective treatment with tolerable toxicity for vaginal recurrences of cervical cancer in patients with prior surgery. Recurrence pattern and irradiation history were important prognostic factors.

Classification of Chaotic Squeak And Rattle Vibrations By CNN Using Recurrence Pattern

The chaotic squeak and rattle (S&R) vibrations in mechanical systems were classified by deep learning. The rattle, single-mode, and multi-mode squeak models were constructed to generate chaotic S&R signals. The repetition of nonlinear signals generated by them was visualized using an unthresholded recurrence plot and learned using a convolutional neural network (CNN). The results showed that even if the signal of the S&R model is chaos, it could be classified. The accuracy of the classification was verified by calculating the Lyapunov exponent of the vibration signal. The numerical experiment confirmed that the CNN classification using nonlinear vibration images as the proposed procedure has more than 90% accuracy. The chaotic status and each model can be classified into six classes.