<p>Medical image fusion is crucial for various applications, yet existing fusion models often encounter issues such as low quality, information loss, and insufficient contrast. This paper presents a novel approach to medical image fusion that directly addresses these challenges by enhancing brightness and contrast, improving overall image quality, and preserving essential structural features. Our method begins by decomposing input images into a base component and two detail components using a three-component decomposition (TCD) method with a truncated Huber filter (THF). The detail components are subsequently fused using an integrated approach combining the structure tensor saliency detection operator (STSDO), distance-weighted regional energy (DWRE), fast guided filter (FGF), and coupled neural P system (CNPS). The base components are then fused through an adaptive method guided by the moth-flame optimization (MFO) algorithm. To evaluate our approach, we used a dataset of 120 pairs of medical images and assessed performance with metrics such as <InlineEquation ID="IEq1"> <InlineMediaObject> <ImageObject Color="BlackWhite" FileRef="521_2025_11072_Article_IEq1.gif" Format="GIF" Height="17" Rendition="HTML" Resolution="72" Type="Linedraw" Width="44" /> </InlineMediaObject> <EquationSource Format="TEX">$Q_{MLI}$</EquationSource> <EquationSource Format="MATHML"><math> <msub> <mi>Q</mi> <mrow> <mi mathvariant="italic">MLI</mi> </mrow> </msub> </math></EquationSource> </InlineEquation>, <InlineEquation ID="IEq2"> <InlineMediaObject> <ImageObject Color="BlackWhite" FileRef="521_2025_11072_Article_IEq2.gif" Format="GIF" Height="17" Rendition="HTML" Resolution="72" Type="Linedraw" Width="33" /> </InlineMediaObject> <EquationSource Format="TEX">$Q_{CI}$</EquationSource> <EquationSource Format="MATHML"><math> <msub> <mi>Q</mi> <mrow> <mi mathvariant="italic">CI</mi> </mrow> </msub> </math></EquationSource> </InlineEquation>, <InlineEquation ID="IEq3"> <InlineMediaObject> <ImageObject Color="BlackWhite" FileRef="521_2025_11072_Article_IEq3.gif" Format="GIF" Height="17" Rendition="HTML" Resolution="72" Type="Linedraw" Width="36" /> </InlineMediaObject> <EquationSource Format="TEX">$Q_{AG}$</EquationSource> <EquationSource Format="MATHML"><math> <msub> <mi>Q</mi> <mrow> <mi mathvariant="italic">AG</mi> </mrow> </msub> </math></EquationSource> </InlineEquation>, <InlineEquation ID="IEq4"> <InlineMediaObject> <ImageObject Color="BlackWhite" FileRef="521_2025_11072_Article_IEq4.gif" Format="GIF" Height="20" Rendition="HTML" Resolution="72" Type="Linedraw" Width="50" /> </InlineMediaObject> <EquationSource Format="TEX">$Q^{AB/F}$</EquationSource> <EquationSource Format="MATHML"><math> <msup> <mi>Q</mi> <mrow> <mi>A</mi> <mi>B</mi> <mo stretchy="false">/</mo> <mi>F</mi> </mrow> </msup> </math></EquationSource> </InlineEquation>, <InlineEquation ID="IEq5"> <InlineMediaObject> <ImageObject Color="BlackWhite" FileRef="521_2025_11072_Article_IEq5.gif" Format="GIF" Height="17" Rendition="HTML" Resolution="72" Type="Linedraw" Width="26" /> </InlineMediaObject> <EquationSource Format="TEX">$Q_{P}$</EquationSource> <EquationSource Format="MATHML"><math> <msub> <mi>Q</mi> <mi>P</mi> </msub> </math></EquationSource> </InlineEquation>, <InlineEquation ID="IEq6"> <InlineMediaObject> <ImageObject Color="BlackWhite" FileRef="521_2025_11072_Article_IEq6.gif" Format="GIF" Height="17" Rendition="HTML" Resolution="72" Type="Linedraw" Width="29" /> </InlineMediaObject> <EquationSource Format="TEX">$Q_{W}$</EquationSource> <EquationSource Format="MATHML"><math> <msub> <mi>Q</mi> <mi>W</mi> </msub> </math></EquationSource> </InlineEquation>, <InlineEquation ID="IEq7"> <InlineMediaObject> <ImageObject Color="BlackWhite" FileRef="521_2025_11072_Article_IEq7.gif" Format="GIF" Height="17" Rendition="HTML" Resolution="72" Type="Linedraw" Width="27" /> </InlineMediaObject> <EquationSource Format="TEX">$Q_{E}$</EquationSource> <EquationSource Format="MATHML"><math> <msub> <mi>Q</mi> <mi>E</mi> </msub> </math></EquationSource> </InlineEquation>, and <InlineEquation ID="IEq8"> <InlineMediaObject> <ImageObject Color="BlackWhite" FileRef="521_2025_11072_Article_IEq8.gif" Format="GIF" Height="17" Rendition="HTML" Resolution="72" Type="Linedraw" Width="36" /> </InlineMediaObject> <EquationSource Format="TEX">$Q_{CB}$</EquationSource> <EquationSource Format="MATHML"><math> <msub> <mi>Q</mi> <mrow> <mi mathvariant="italic">CB</mi> </mrow> </msub> </math></EquationSource> </InlineEquation>. Experimental results demonstrate that the proposed model significantly enhances brightness and contrast and more effectively preserves information compared to recent methods.</p>

An effective medical image fusion method utilizing moth-flame optimization and coupled neural P systems

期刊论文
发表日期：

2025年3月11日

查看全文

Phu-Hung Dinh,
Thi-Hong-Ha Le,
Nguyen Long Giang

摘要

Medical image fusion is crucial for various applications, yet existing fusion models often encounter issues such as low quality, information loss, and insufficient contrast. This paper presents a novel approach to medical image fusion that directly addresses these challenges by enhancing brightness and contrast, improving overall image quality, and preserving essential structural features. Our method begins by decomposing input images into a base component and two detail components using a three-component decomposition (TCD) method with a truncated Huber filter (THF). The detail components are subsequently fused using an integrated approach combining the structure tensor saliency detection operator (STSDO), distance-weighted regional energy (DWRE), fast guided filter (FGF), and coupled neural P system (CNPS). The base components are then fused through an adaptive method guided by the moth-flame optimization (MFO) algorithm. To evaluate our approach, we used a dataset of 120 pairs of medical images and assessed performance with metrics such as $Q_{MLI}$ $Q_{MLI}$ , $Q_{CI}$ $Q_{CI}$ , $Q_{AG}$ $Q_{AG}$ , $Q^{AB/F}$ $Q^{A B / F}$ , $Q_{P}$ $Q_{P}$ , $Q_{W}$ $Q_{W}$ , $Q_{E}$ $Q_{E}$ , and $Q_{CB}$ $Q_{CB}$ . Experimental results demonstrate that the proposed model significantly enhances brightness and contrast and more effectively preserves information compared to recent methods.