I'm trying to generate properly a table with variable and associated p-value with kable package in a quarto document.I use kable and kableExtra packages.My yaml header is the following one:
---title: "Towards quantification of image quality in chest CT clinical images : part I. Extraction of imaging features for prediction of system performances and lesion detectability"author: - name: F. Gardavaud orcid: 0000-0001-9767-3241 corresponding: true email: francois.gardavaud@aphp.fr roles: - Image database designing - Article writing - Radiomics computing - Data analysis affiliations: - APHP - name: Hugo Pasquier orcid: 0000-0001-9823-3563 corresponding: false roles: - Radiomics computing - Data analysis - Article reading affiliations: - GE HealthCaredate: last-modifiedkeywords: - Computed Tomography - Radiomics - Image quality metricsabstract: | Introduction: Task-based CT image quality is commonly assessed using large homogenous Regions Of Interests (ROI). Identifying such regions within chest images being almost impossible, the purpose of this study was to evaluate the capability to predict the system performances and the lesion detectability based on small ROIs.<br> \newline Material and methods:<br> \newline Conclusions:plain-language-summary: Bli bli bikey-points: - Image quality metrics based on gold standard and radiomics features have been performed on a standard control quality phantom, - Radiomics features, computed in small analysis ROI, could be used to replace image quality metrics in computed tomography examination, - Radiomics features could be used to assess the image quality in computed tomography clinical image.number-sections: truehighlight-style: pygmentsformat: html: code-fold: true html-math-method: katex embed-resources: true self-contained-math: true df-print: "kable" pdf: toc: false df-print: "kable" number-sections: true colorlinks: true geometry: - top=30mm - left=30mm docx: defaulteditor: visualbibliography: references.bibcsl: physica-medica.csl---Here's my code (just the corresponding chunk):
{r}#| code-summary: "Kruskal-Wallis and Dunn post-hoc tests for the NPS, FD, TTF and d' values"#| message: false#| warning: false#| echo: false#| label: tbl-res-p-value#| tbl-cap: p values of the Kruskal-Wallis and Dunn post-hoc comparison tests of the average spatial frequency of the Noise Power Spectrum f<sub>av</sub>, the spatial frequency at which the Task Transfer Function was reduced by 50% (TTF<sub>50</sub>) and detectability indexes (d') between the different CT units. p-value < 0.05 were considered significant. \\*, \\**, \\*** notifications indicate p < 0.05, p < 0.01, p < 0.001 respectively.#| cache: true # pour ne pas recalculer les résultats à chaque compilation#| results: 'asis'# comme la normalité b'est pas respecté, on exécute le test de Kruskal-Wallis pour tester l'égalité des moyennes des variables NPS, FD, TTF et d' entre les 3 modèles de CT# pour le NPSkruskal_NPS <- kruskal.test(meanFreq_128 ~ CT, data = DF_IQ_NPS_3CT_selected_rfe)# pour les TTF# pour bonekruskal_TTF_bone <- kruskal.test(f50Bone ~ CT, data = DF_IQ_TTF_3CT_filtered)# pour airkruskal_TTF_air <- kruskal.test(f50Air ~ CT, data = DF_IQ_TTF_3CT_filtered)# pour HEIodinekruskal_TTF_iodine <- kruskal.test(f50HEIodine ~ CT, data = DF_IQ_TTF_3CT_filtered)# pour Phantomkruskal_TTF_edge <- kruskal.test(f50Phantom ~ CT, data = DF_IQ_TTF_3CT_filtered)# pour le dprimekruskal_dprime <- kruskal.test(dPrime ~ CT, data = DF_IQ_dprime_3CT_selected_rfe)# pour le dprime GGNkruskal_dprime_GGN <- kruskal.test(dPrime ~ CT, data = DF_IQ_dprime_3CT_GGN)# pour le dprime partialkruskal_dprime_partial <- kruskal.test(dPrime ~ CT, data = DF_IQ_dprime_3CT_partial)# pour le dprime solidkruskal_dprime_solid <- kruskal.test(dPrime ~ CT, data = DF_IQ_dprime_3CT_solid)# les tests sont tous significatifs, on peut donc faire des tests post-hoc pour déterminer les différences entre les groupes# pour le NPSdunn_NPS <- dunnTest(meanFreq_128 ~ CT, data = DF_IQ_NPS_3CT_selected_rfe, method = "bon")# pour les TTF# pour bonedunn_TTF_bone <- dunnTest(f50Bone ~ CT, data = DF_IQ_TTF_3CT_filtered, method = "holm")# pour airdunn_TTF_air <- dunnTest(f50Air ~ CT, data = DF_IQ_TTF_3CT_filtered, method = "holm")# pour iodinedunn_TTF_iodine <- dunnTest(f50HEIodine ~ CT, data = DF_IQ_TTF_3CT_filtered, method = "holm")# pour phantomdunn_TTF_edge <- dunnTest(f50Phantom ~ CT, data = DF_IQ_TTF_3CT_filtered, method = "holm")# pour le dprimedunn_dprime <- dunnTest(dPrime ~ CT, data = DF_IQ_dprime_3CT_selected_rfe, method = "holm")# pour le dprime GGNdunn_dprime_GGN <- dunnTest(dPrime ~ CT, data = DF_IQ_dprime_3CT_GGN, method = "holm")# pour le dprime partialdunn_dprime_partial <- dunnTest(dPrime ~ CT, data = DF_IQ_dprime_3CT_partial, method = "holm")# pour le dprime soliddunn_dprime_solid <- dunnTest(dPrime ~ CT, data = DF_IQ_dprime_3CT_solid, method = "holm")# on créé un dataframe pour stocker les résultats des tests de Kruskal-Wallis et des tests post-hoc# pour ce faire la première colone correspond aux variables, la seconde aux p-values de Kruskal-Wallis. La troisième, quatrième est cinquième colonne correspondent aux p-values des tests post-hoc de Dunn avec en titre de colonne la comparaison entre les groupesDF_kruskal_dunn <- data.frame( # Metric = c("$\\mathrm{f_{av}}$", "$\\mathrm{TTF_{50\\_Bone}}$", "$\\mathrm{TTF_{50\\_Air}}$", "$\\mathrm{TTF_{50\\_Iodine}}$", "$\\mathrm{TTF_{50\\_Edge}}$", "$\\mathrm{d'_{Solid}}$", "$\\mathrm{d'_{Partial}}$", "$\\mathrm{d'_{GGN}}$"), # Metric = c("x","a","b","c","d","e","f","g"), Metric = c("$\\mathrm{f_{av}}$", "$\\mathrm{TTF_{50\\_Bone}}$", "$\\mathrm{TTF_{50\\_Air}}$", "$\\mathrm{TTF_{50\\_Iodine}}$", "$\\mathrm{TTF_{50\\_Edge}}$", "$\\mathrm{d'_{Solid}}$", "$\\mathrm{d'_{Partial}}$", "$\\mathrm{d'_{GGN}}$"), `Kruskal.Wallis` = c(kruskal_NPS$p.value, kruskal_TTF_bone$p.value, kruskal_TTF_air$p.value, kruskal_TTF_iodine$p.value, kruskal_TTF_edge$p.value, kruskal_dprime_solid$p.value, kruskal_dprime_partial$p.value, kruskal_dprime_GGN$p.value), Canon_vs_GE = c(dunn_NPS$res$P.adj[1], dunn_TTF_bone$res$P.adj[1], dunn_TTF_air$res$P.adj[1], dunn_TTF_iodine$res$P.adj[1], dunn_TTF_edge$res$P.adj[1], dunn_dprime_solid$res$P.adj[1], dunn_dprime_partial$res$P.adj[1], dunn_dprime_GGN$res$P.adj[1]), Canon_vs_Siemens = c(dunn_NPS$res$P.adj[2], dunn_TTF_bone$res$P.adj[2], dunn_TTF_air$res$P.adj[2], dunn_TTF_iodine$res$P.adj[2], dunn_TTF_edge$res$P.adj[2], dunn_dprime_solid$res$P.adj[2], dunn_dprime_partial$res$P.adj[2], dunn_dprime_GGN$res$P.adj[2]), GE_vs_Siemens = c(dunn_NPS$res$P.adj[3], dunn_TTF_bone$res$P.adj[3], dunn_TTF_air$res$P.adj[3], dunn_TTF_iodine$res$P.adj[3], dunn_TTF_edge$res$P.adj[3], dunn_dprime_solid$res$P.adj[3], dunn_dprime_partial$res$P.adj[3], dunn_dprime_GGN$res$P.adj[3]))# on met les valeurs en format décimal fixe pour les valeurs de p > 0.001 et on ajoute des * pour les p-values < 0.05, ** pour les p-values < 0.01 et *** pour les p-values < 0.001DF_kruskal_dunn <- DF_kruskal_dunn %>% mutate(across(c(Kruskal.Wallis, Canon_vs_GE, Canon_vs_Siemens, GE_vs_Siemens), ~ ifelse(. < 0.001, paste0(formatC(., format = "f", digits = 3), "***"), ifelse(. < 0.01, paste0(formatC(., format = "f", digits = 3), "**"), ifelse(. < 0.05, paste0(formatC(., format = "f", digits = 3), "*"), formatC(., format = "f", digits = 3))))))# pour le colonnes p, Canon_vs_GE, Canon_vs_Siemens et GE_vs_Siemens on passe du format puissance au format numérique et on remplace les valeurs inférieures à 0.001 par "< 0.001"DF_kruskal_dunn <- DF_kruskal_dunn %>% mutate(across(c(Kruskal.Wallis, Canon_vs_GE, Canon_vs_Siemens, GE_vs_Siemens), ~ ifelse(. < 0.001, "< 0.001 ***", formatC(., format = "f", digits = 3))))# Renommer la colonne Kruskal.Wallis en Kruskal WallisDF_kruskal_dunn <- DF_kruskal_dunn %>% rename(`Kruskal Wallis` = Kruskal.Wallis)# on affiche le dataframekable( DF_kruskal_dunn, escape = FALSE, booktabs = TRUE ) # kable_styling(full_width = TRUE) # kable_styling(latex_options = c("striped", "scale_down"))# %>% #add_header_above(c(" " = 2, "Dunn post-hoc" = 3), italic = TRUE)In this case the table is generated correctly but if I'm trying to uncomment add_header_above(c(" " = 2, "Dunn post-hoc" = 3), italic = TRUE)the first column which contains Latex caracters is displayed with all the characters, i.e. $\\mathrm{f_{av}}$ for the first value instead of the right format.
For test purpose here's a reprex :
DF_kruskal_dunn <- data.frame( Metric = c("$\\mathrm{f_{av}}$", "$\\mathrm{TTF_{50\\_Bone}}$", "$\\mathrm{TTF_{50\\_Air}}$", "$\\mathrm{TTF_{50\\_Iodine}}$", "$\\mathrm{TTF_{50\\_Edge}}$", "$\\mathrm{d'_{Solid}}$", "$\\mathrm{d'_{Partial}}$", "$\\mathrm{d'_{GGN}}$"), `Kruskal.Wallis` = c(0.4, 0.0002, 0.3, 0.005, 0.3, 0.0001, 0.0001, 0.0001), Canon_vs_GE = c(0.4, 0.0002, 0.3, 0.5, 0.003, 0.0001, 0.0001, 0.0001), Canon_vs_Siemens = c(0.4, 0.0002, 0.3, 0.5, 0.3, 0.0001, 0.00003, 0.000001), GE_vs_Siemens = c(0.4, 0.0002, 0.3, 0.5, 0.3, 0.0001, 0.0002, 0.00001))# we put the values in fixed decimal format for p-values > 0.001 and add * for p-values < 0.05, ** for p-values < 0.01 and *** for p-values < 0.001.DF_kruskal_dunn <- DF_kruskal_dunn %>% mutate(across(c(Kruskal.Wallis, Canon_vs_GE, Canon_vs_Siemens, GE_vs_Siemens), ~ ifelse(. < 0.001, paste0(formatC(., format = "f", digits = 3), "***"), ifelse(. < 0.01, paste0(formatC(., format = "f", digits = 3), "**"), ifelse(. < 0.05, paste0(formatC(., format = "f", digits = 3), "*"), formatC(., format = "f", digits = 3))))))# for the columns p, Canon_vs_GE, Canon_vs_Siemens and GE_vs_Siemens we switch from power format to numeric format and replace values less than 0.001 by “< 0.001”.DF_kruskal_dunn <- DF_kruskal_dunn %>%DF_kruskal_dunn <- DF_kruskal_dunn %>%mutate(across(c(Kruskal.Wallis, Canon_vs_GE, Canon_vs_Siemens, GE_vs_Siemens), ~ ifelse(. < 0.001, "< 0.001 ***", formatC(., format = "f", digits = 3)))) # Rename columns with spaces for greater clarityDF_kruskal_dunn <- DF_kruskal_dunn %>% rename(`Kruskal Wallis` = Kruskal.Wallis)DF_kruskal_dunn <- DF_kruskal_dunn %>% rename(`Canon vs GE` = Canon_vs_GE)DF_kruskal_dunn <- DF_kruskal_dunn %>% rename(`Canon vs Siemens` = Canon_vs_Siemens)DF_kruskal_dunn <- DF_kruskal_dunn %>% rename(`GE vs Siemens` = GE_vs_Siemens)# dataframe diasplaykable( DF_kruskal_dunn, escape = FALSE, booktabs = TRUE)# kable_styling(full_width = TRUE)# kable_styling(latex_options = c("striped", "scale_down"))# %>%# add_header_above(c(" " = 2, "Dunn post-hoc" = 3), italic = TRUE)Could you help me to solve this problem?