<p>In this article, a novel deep neural network architecture is presented for distinguishing subjects of Parkinson’s Disease (PD) from Cognitively Normal (CN) cohort. The proposed architecture combines the representation learning capacities and discriminating capabilities of convolutional neural network to improve classification performance of the network. For this purpose, the network consists of a classification sub-network and multiple reconstruction sub-networks. The classification sub-network discriminates between PD and CN subjects and the reconstruction sub-networks are used to regularize the individual layers of the classification sub-network. All sub-networks are trained simultaneously to optimize the customized objective function of the network. The network is trained on Single Photon Emission Computerized Tomography (SPECT) images obtained from baseline study of Parkinson’s Progression Marker Initiative (PPMI). The proposed approach provides classification accuracy, recall, precision and F1-score of <InlineEquation ID="IEq1"> <InlineMediaObject> <ImageObject Color="BlackWhite" FileRef="11042_2025_20661_Article_IEq1.gif" Format="GIF" Height="19" Rendition="HTML" Resolution="72" Type="Linedraw" Width="105" /> </InlineMediaObject> <EquationSource Format="TEX">\(97.14(\pm 0.63)\%\)</EquationSource> <EquationSource Format="MATHML"><math> <mrow> <mn>97.14</mn> <mo stretchy="false">(</mo> <mo>±</mo> <mn>0.63</mn> <mo stretchy="false">)</mo> <mo>%</mo> </mrow> </math></EquationSource> </InlineEquation>, <InlineEquation ID="IEq2"> <InlineMediaObject> <ImageObject Color="BlackWhite" FileRef="11042_2025_20661_Article_IEq2.gif" Format="GIF" Height="19" Rendition="HTML" Resolution="72" Type="Linedraw" Width="105" /> </InlineMediaObject> <EquationSource Format="TEX">\(95.89(\pm 1.25)\%\)</EquationSource> <EquationSource Format="MATHML"><math> <mrow> <mn>95.89</mn> <mo stretchy="false">(</mo> <mo>±</mo> <mn>1.25</mn> <mo stretchy="false">)</mo> <mo>%</mo> </mrow> </math></EquationSource> </InlineEquation>,<InlineEquation ID="IEq3"> <InlineMediaObject> <ImageObject Color="BlackWhite" FileRef="11042_2025_20661_Article_IEq3.gif" Format="GIF" Height="19" Rendition="HTML" Resolution="72" Type="Linedraw" Width="105" /> </InlineMediaObject> <EquationSource Format="TEX">\(95.01(\pm 2.51)\%\)</EquationSource> <EquationSource Format="MATHML"><math> <mrow> <mn>95.01</mn> <mo stretchy="false">(</mo> <mo>±</mo> <mn>2.51</mn> <mo stretchy="false">)</mo> <mo>%</mo> </mrow> </math></EquationSource> </InlineEquation>, and <InlineEquation ID="IEq4"> <InlineMediaObject> <ImageObject Color="BlackWhite" FileRef="11042_2025_20661_Article_IEq4.gif" Format="GIF" Height="19" Rendition="HTML" Resolution="72" Type="Linedraw" Width="129" /> </InlineMediaObject> <EquationSource Format="TEX">\(0.9542(\pm 0.0093)\%\)</EquationSource> <EquationSource Format="MATHML"><math> <mrow> <mn>0.9542</mn> <mo stretchy="false">(</mo> <mo>±</mo> <mn>0.0093</mn> <mo stretchy="false">)</mo> <mo>%</mo> </mrow> </math></EquationSource> </InlineEquation> respectively. When compared to recent state of the art methods for PD detection, the present work reports upto <InlineEquation ID="IEq5"> <InlineMediaObject> <ImageObject Color="BlackWhite" FileRef="11042_2025_20661_Article_IEq5.gif" Format="GIF" Height="16" Rendition="HTML" Resolution="72" Type="Linedraw" Width="31" /> </InlineMediaObject> <EquationSource Format="TEX">\(10\%\)</EquationSource> <EquationSource Format="MATHML"><math> <mrow> <mn>10</mn> <mo>%</mo> </mrow> </math></EquationSource> </InlineEquation> improvement in the classification accuracy. The proposed network also employs Gaussian noise to reduce the overfitting by randomly augmenting the training samples.</p>

错误:搜索内容不能为空,请输入英文关键词
错误:关键词超出字数限制,请精简
高级检索

A deep learning approach for classification of Parkinson’s disease using single photon emission computerized tomography

  • Reema Ganotra,
  • Shailender Gupta

摘要

In this article, a novel deep neural network architecture is presented for distinguishing subjects of Parkinson’s Disease (PD) from Cognitively Normal (CN) cohort. The proposed architecture combines the representation learning capacities and discriminating capabilities of convolutional neural network to improve classification performance of the network. For this purpose, the network consists of a classification sub-network and multiple reconstruction sub-networks. The classification sub-network discriminates between PD and CN subjects and the reconstruction sub-networks are used to regularize the individual layers of the classification sub-network. All sub-networks are trained simultaneously to optimize the customized objective function of the network. The network is trained on Single Photon Emission Computerized Tomography (SPECT) images obtained from baseline study of Parkinson’s Progression Marker Initiative (PPMI). The proposed approach provides classification accuracy, recall, precision and F1-score of \(97.14(\pm 0.63)\%\) 97.14 ( ± 0.63 ) % , \(95.89(\pm 1.25)\%\) 95.89 ( ± 1.25 ) % , \(95.01(\pm 2.51)\%\) 95.01 ( ± 2.51 ) % , and \(0.9542(\pm 0.0093)\%\) 0.9542 ( ± 0.0093 ) % respectively. When compared to recent state of the art methods for PD detection, the present work reports upto \(10\%\) 10 % improvement in the classification accuracy. The proposed network also employs Gaussian noise to reduce the overfitting by randomly augmenting the training samples.